<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Slingshot on Medium]]></title>
        <description><![CDATA[Stories by Slingshot on Medium]]></description>
        <link>https://medium.com/@heyslingshot?source=rss-cc8458c4630f------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*51Zq5XFkO2Npsf4wyYFqoQ.png</url>
            <title>Stories by Slingshot on Medium</title>
            <link>https://medium.com/@heyslingshot?source=rss-cc8458c4630f------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 03:33:14 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@heyslingshot/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[4 Reasons Your AI Coding Agent Is Only as Good as Your Ticket]]></title>
            <link>https://medium.com/@heyslingshot/4-reasons-your-ai-coding-agent-is-only-as-good-as-your-ticket-33e84dd4d67a?source=rss-cc8458c4630f------2</link>
            <guid isPermaLink="false">https://medium.com/p/33e84dd4d67a</guid>
            <category><![CDATA[ai-coding]]></category>
            <category><![CDATA[agile]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Slingshot]]></dc:creator>
            <pubDate>Fri, 06 Nov 2026 08:01:09 GMT</pubDate>
            <atom:updated>2026-11-12T16:19:26.290Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Ywv1M1lmQTK3vg4t.png" /></figure><p>Most engineering teams spent the last two years getting faster at writing code. Fewer noticed the constraint quietly moving upstream, out of the build and into the ticket.</p><p>Hand a developer a thin ticket, and nothing breaks. They can ask two questions and build the right thing. Hand the same ticket for an autonomous agent, and it reads what you wrote, fills the gaps on its own, and ships something that looks finished either way.</p><p>That’s what changes once agentic workflows reach real backlogs. Your output stops depending on how well your team codes and starts depending on how completely your team specifies. The job description shifted more than any tool did.</p><h3>Summary</h3><p><em>Teams running coding agents are finding that the fix isn’t better prompts, it’s acceptance criteria that pass or fail cleanly, test scenarios written before the code, and a check that interrogates the ticket before the build starts. Slingshot’s Doug Compton runs a pipeline where a ticket goes in and a review-ready pull request comes out, and the habits that keep it honest cost nothing but attention. What they protect you from are the decisions your software is quietly making without you.</em></p><h3>1. Automation removes the conversation that used to catch your gaps</h3><p>The traditional ticket works because it’s only half the requirement; the other half lives in a hallway, a Slack thread, or a five-minute call.</p><p>“When you submit a ticket, it can be kind of vague. The developer will call you and ask you what you’re talking about, and you can have this back-and-forth conversation,” said Doug Compton, Director of AI Engineering at Slingshot. “But if this is an automated factory, there is no back and forth. The ticket needs to include everything that it needs.”</p><p>Doug builds what he calls a dark factory for code: a manufacturing term for plants so automated they turned off the lights because nobody was left on the floor to need them. People still design the line and handle the exceptions, but the machines run it.</p><p>His software version takes in a ticket and hands out a review-ready pull request. An agent researches the problem, writes a plan, implements it, and validates its own work at every step. A human manages the ticket up front, and a human reviews the pull requests on the backend, so everything in between runs unattended.</p><p>That design is exactly why the front of the line carries so much weight now. Your developers have to answer every clarifying question in advance, in writing, by the person who opened the ticket, and for most teams that person is a business user who has never had to write a requirement that can stand on its own.</p><p>Doug is blunt about what that did to his week: “Most of my job went from typing code to typing tickets.”</p><h3>2. A gap in the ticket doesn’t stop the work, so it becomes an invisible decision</h3><p>This is what catches teams off guard: an incomplete ticket doesn’t produce an error. It produces a wrong answer.</p><p>“If there’s missing information, the AI will fill it in with something it makes up,” Doug said. “If it does something stupid, you can call it a hallucination, but if it does something smart, you might not even notice it.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*5F2o53wlH2StQ7L5" /></figure><p>Sit with the second half of that. You catch the obvious failures, but you never catch the reasonable-sounding guesses that quietly became product behavior, because nobody watches the middle of the process to see the decision get made.</p><p>Doug gave an example of a ticket he broke on purpose. He asked AI for a move counter so players could see how many clicks it took them to finish a puzzle. “But once you figure out a puzzle, if you keep clicking, what should it do? We never told it that,” he said. “Should it stop once you solve it? Or if you click again, will the move counter increase?”</p><p>The factory never asked. “I’ve run this a couple of times, and it’s actually implemented both ways,” Doug said. Nothing failed either time. Every validator passed, every test went green, and the agent made a product decision on behalf of the business and recorded it as a fact.</p><p>Scale that across a quarter of backlog, and you’re no longer looking at a quality problem, because small, undocumented decisions start accumulating inside shipped software with no author and no paper trail.</p><h3>3. Acceptance criteria are the contract, and “sort of” is not a valid result</h3><p>When tickets are this load-bearing, they need structure, not just length. The test Doug recommends is whether another agent, with no context beyond the conversation history, could pick up the plan and build the right thing from it alone.</p><p>That standard produces two rules. Rule #1 protects what the business asked for: “The plan may also add more stuff to it, or find new acceptance criteria, but it should never drop or reword what was in the ticket,” Doug said. An agent can sharpen your requirement, but it cannot quietly trade it for an easier one.</p><p>Rule #2 governs how each criterion gets graded. Doug said: “Each one should have a pass or fail, but never a ‘sort of.’ There needs to be a definitive yes, this passed, or no, this didn’t pass.”</p><p>That one does more work than it looks like. A criterion you can only grade as mostly met is a criterion no automated validator can act on, so it falls back to human review at exactly the point where no human is present. Vague in, vague out.</p><p>Tests are where those criteria stop being prose. Every acceptance criterion includes at least one scenario naming concrete inputs and the exact result they should produce, and edge cases get scenarios too, since the empty string and the negative number are precisely where a confident guess does the most damage. Write each scenario before the fix, watch it fail, and only then let the agent change the code until it passes.</p><p>So run the handoff test on your own backlog. If a stranger with no one to ask couldn’t build it correctly, someone has to fill that gap. Because if nobody does, the agent will.</p><h3>4. The fix is a validator on the ticket itself, not more discipline from your business users</h3><p>The obvious response is to tell everyone to write better tickets, but that won’t work. What if the ticket could help improve itself instead?</p><p>Doug accomplished this with a check at the factory gate. “I created a skill that will look at the ticket, then at the source code, and it will ask me questions,” he said. “It will say things like, ‘I see that this wasn’t mentioned. What should it do?’”</p><p>The ticket gets interrogated before the build starts, so the gaps surface before anything gets built. What is missing or unanswered? What should happen at the edges, and on errors? What is explicitly out of scope? Put the answers in the ticket, not in a chat window where they evaporate.</p><p>Notice what this preserves. The AI surfaces the gaps because it reads the entire codebase in seconds and spots what the requirement doesn’t cover, but the human decides what fills them. Only the human knows what the business actually wants; a developer controls what makes it into the ticket.</p><p>That split explains why he stays in the loop, even though agents now write most of the code: “The AI is just as good at writing code as I am, but it still doesn’t know the context of your business process.”</p><p>The same logic extends to the codebase itself. Doug wrote another skill that generates documentation specifically for an AI to read, and it lands in a predictable structure and gets indexed where the agent will always see it. He spends more time on research and architecture than he used to, and he writes it down somewhere the machine can find it. The context that lives only in a senior developer’s head is context the factory can’t use.</p><p>One more guardrail belongs in this category, since it’s also a specification problem: bound the loop. Doug caps his factory at three rounds, and if it can’t converge, it writes an escalation file explaining why it stopped. A loop that never resolves doesn’t fail in any visible way, so it will keep spending your money until somebody notices, most likely too late.</p><h3>What this actually changes</h3><p>The factory works, and that’s exactly why this matters. Tickets go in, validated pull requests come out, and the tests catch regressions a human reviewer would’ve skimmed past. The machinery isn’t the weak link anymore.</p><p>What changes is where the leverage sits. When agents write the code, the quality ceiling of your software gets set by the clarity of your requirements, the completeness of your acceptance criteria, and the quality of the documentation your systems can genuinely read. Those aren’t developer problems, so they belong to the people who run the process.</p><p>The question for your org is not whether your team can use coding agents, because they probably already do. The question is whether your intake would survive losing the hallway conversation.</p><p>Pull your last ten tickets and read each one as if the person who wrote it has gone silent and will never answer another question about it. Every gap you find is a decision you have already delegated to something that will make it confidently, quietly, and without telling you.</p><h3>Scope Before You Connect</h3><p><a href="https://www.yslingshot.com/ai-tool-questions/">6 Questions To Answer Before You Turn On Any AI Tool ⇢</a></p><h3>Written by: Savannah Cherry</h3><p>Savannah leads marketing and new business at Slingshot. She writes, posts, and creates all things Slingshot, and helps companies navigate working with a tech partner for the first time. While she isn’t developing software, her CIS minor and a tendency to tinker with AI tools to streamline her own work keep her up to speed on the team’s work. She co-organizes and hosts the Louisville AI Exchange, and she can’t rest until all her work is done.</p><p><a href="https://www.yslingshot.com/author/savannah/">View All Posts</a></p><h3>Expert: Doug Compton</h3><p>Born and raised in Louisville, Doug’s interest in technology started at 11 when he began writing computer games. What began as a hobby turned into his career. With broad interests that range anywhere from snorkeling, science, WWII history and real estate, Doug uses his “down time“ to create new technologies for mobile and web applications.</p><p><a href="https://www.linkedin.com/in/doug-compton-1564438/">Linkedin</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=33e84dd4d67a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[21 Years of Reinvention, and Slingshot Isn’t Done]]></title>
            <link>https://medium.com/@heyslingshot/21-years-of-reinvention-and-slingshot-isnt-done-8e79adfbf6db?source=rss-cc8458c4630f------2</link>
            <guid isPermaLink="false">https://medium.com/p/8e79adfbf6db</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[business-strategy]]></category>
            <category><![CDATA[leadership]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[software-development]]></category>
            <dc:creator><![CDATA[Slingshot]]></dc:creator>
            <pubDate>Fri, 30 Oct 2026 07:01:34 GMT</pubDate>
            <atom:updated>2026-11-12T16:19:04.452Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*9w49IbRMjYQRo1WU.png" /></figure><p>Most companies treat change as something that happens to them. Slingshot treats it as the job.</p><p>That distinction explains much about the last 21 years, and what’s happening right now. David Galownia, who started the company in 2005, has moved away from the day-to-day. Chris Howard, who joined in 2017, is officially President and is now running the business and setting its direction.</p><p>It’s easy to frame that as an ending; it’s closer to the opposite. Here’s how Slingshot got here, what’s actually changing, and what isn’t.</p><h3>Summary</h3><p><em>Slingshot’s answer to two decades of market upheaval turns out to be a skill the company practices on purpose, not a reaction it scrambles into. A near-fatal pivot a dozen years ago rebuilt the business from a single-technology dev shop into a full solution provider, and the same instinct now drives an AI shift that rewrote job descriptions, titles, and roles across every function, plus two new offerings aimed at where clients are actually stuck. David Galownia and Chris Howard each explain what the handoff changes, what it deliberately doesn’t, and why the only promise that has held since 2005 is the one about people.</em></p><h3>Change Is the Reason We’re Still Here</h3><p>Start with 2005, because the market looked nothing like it does today.</p><p>“There were no mobile apps. There was no AI,” David said. “Custom programming was a big thing, and there weren’t a lot of SaaS tools either. So there was a lot more custom development, and a lot more internal development.”</p><p>The company’s idea was simple enough to survive it all. Find people who are genuinely passionate and genuinely talented, put them together, and build the right things the right way.</p><p>“The market has changed dramatically,” David said. “But the fundamental idea of Slingshot, hiring really talented people, creating a positive environment, and letting them use that passion to build great things, isn’t any different.”</p><p>The philosophy held still while everything around it moved. Smartphones became a thing. Design stopped being a finishing touch and became a requirement. SaaS swallowed work companies used to build in-house. And now AI is rewriting the category again. Through every one of those shifts, our promise to clients stayed the same: help them get better with technology, whatever it is that year.</p><h3>The Pivot That Almost Didn’t Happen</h3><p>Ask David what he’s proudest of, and he doesn’t name a product launch. He names the hardest stretch the company ever had.</p><p>Roughly a dozen years ago, Slingshot was a laser-focused custom development shop built around a single technology. The market moved faster than the company did. Work slowed, revenue dropped significantly, and Slingshot had to let people go. That was the point where the company had to decide what it would be.</p><p>“We could have kept going down that road. We could have doubled down on it,” David said. “But we realized the market didn’t need a team of software development specialists. It needed a full solution provider.”</p><p>So Slingshot rebuilt itself: new skills, new services, new office, new branding, new name. Real investment in design and marketing for the first time.</p><p>“We could have given up at that point, and we didn’t,” David said. “I’m proud that we had a big moment and we were able to pivot definitively and move quickly, and not just persist but flourish.”</p><p>That’s the part worth sitting with. The proudest moment in 21 years wasn’t a win. It was deciding not to quit while rebuilding from the inside out. Companies that treat reinvention as an emergency rarely get good at it. Companies that treat it as a skill get second and third acts.</p><p>Asked whether he would change any of it, David stood firm. “Looking back is valuable, but the experiences we went through solidified the direction we needed to be on,” he said. “Maybe we could have noticed some things sooner. We were a very culture-centered company, which is good, but culture has to meet the market.”</p><p>The answer underneath that is no, he wouldn’t redo anything. “I don’t know that I’d make decisions any differently,” David said. “I’m happy we got where we are today, and that it wasn’t necessarily a straight line.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*1QIxFeULEedBXa6u" /></figure><h3>What Chris Saw in 2017</h3><p>Chris joined during that rebuild, after spending most of his career on the road between California, Texas, and India, learning new ways to build products and wanting to be home more.</p><p>“What sold me on Slingshot was David, 100 percent,” Chris said. “I just felt like we could do something good together.”</p><p>He’s honest about what came next. “Those first couple years, everything about it was challenging. We were redoing everything,” Chris said. “I probably underestimated how hard that would be. But I wouldn’t change it; it had to be that way.”</p><p>The surprise was how strongly embedded the company already was locally. “I was so disconnected from Louisville at that point. I didn’t know what businesses were here; I didn’t know anybody,” Chris said. “And a lot of those customers are still with Slingshot today, ten years later.”</p><p>That’s the quiet proof point in a transition similar to this one. Continuity isn’t a press-release promise. It shows up in relationships that outlast org charts.</p><h3>Why Now, and Why Chris</h3><p>David isn’t leaving. He still owns the company and has no plans to sell it. What changed is who makes the final call.</p><p>Part of that is pull. Slingshot Ventures, started five years ago, co-founds companies with operators who come through Slingshot’s orbit. One of them, <a href="https://p3-advantage.com/">P3 Technology</a>, took off, and it gave David something he’d been missing.</p><p>“That scratched part of an itch I had to be more involved on the product side versus always the business side,” he said.</p><p>The bigger part is a judgment about what the company needs next. “Slingshot has been going strong for over 20 years, and we’re still growing,” David said. “It’s a good time to let somebody do some new things with it, and have the freedom to do so without me being a bottleneck.”</p><p>Then David made the case for Chris in two parts. The track record: ten years of bringing new things into the company, new ways of building software, the processes around them, design and design thinking, and an early and aggressive read on AI that put Slingshot ahead of the wave rather than behind it.</p><p>And the culture fit, which David treats as the higher bar. “Slingshot’s culture is so critical, and Chris is trusted and respected by the team and by me,” he said. “Chris and I get along well, but we’re also different in a lot of ways. I think Slingshot is going to do some new things because of that, and I’m happy to see where that goes.”</p><p>That last line matters. This isn’t a handoff designed to keep everything identical. It’s a handoff designed to let the company change again.</p><p>None of this is new ground for Chris. He already knows most of Slingshot’s clients, and he and Dave have worked together for years on the company’s future. What’s different is where he spends his time now. Most of it goes to one question: where Slingshot fits in the market next.</p><h3>AI Changed Every Role, Not Just the Technical Ones</h3><p>Ask Chris what Slingshot had to change internally to become genuinely AI-driven, and the answer is short: “Everything,” he said. “Every role within Slingshot has been impacted in some way.”</p><p>The change that mattered most wasn’t buying a tool. It was permission to step away from billable work, and Dave approved it a couple of years ago.</p><p>“Years ago, we started creating time for our key people to learn and explore,” Chris said. “This stuff is moving so rapidly that you have to invest the time, get out there, and see what’s possible.”</p><p>From there it reached the org chart. “We’ve reworked job descriptions, some of the roles within our company have changed, and we’ve even updated job titles,” Chris said. “Developers, designers, and business-side; all of those roles have been challenged. We’ve had to rethink everything about how we work.”</p><p>For a consultancy, he treats the investment as non-negotiable. “We continue to invest time and money into guaranteeing our team are experts in artificial intelligence,” Chris said. “Our customers rely on us to be experts in that space, and to help them find the way.”</p><p>That’s the uncomfortable lesson for leaders watching from the outside. You can’t bolt AI onto a company and leave the roles untouched; the roles are the adoption.</p><h3>Three Years Out: More Diverse, More Collaborative</h3><p>Chris sees a wider company ahead, and a different relationship with clients. “Technology is becoming more and more collaborative,” he said. “In the past we would create the concept and share it with the customer. Now the customer may come to us with the concept.”</p><p>Two offerings anchor the next phase. <a href="https://www.yslingshot.com/launchai/">Launch AI</a> puts Slingshot’s AI experts inside a client’s organization to accelerate adoption through custom training, skills and prompt development, tool selection, and safe usage. When something needs to get built, the same team pulls in Slingshot’s project expertise.</p><p>Orbit AI answers a newer problem. A wave of people are building apps quickly, and then discovering that shipping to production is a different discipline. Orbit AI takes that app and provides the architecture, the cloud setup, and the ongoing guidance on monitoring, security, and maintenance.</p><p>Both offerings point to the same three-year picture. “We’ll be a lot more diverse than where we are today, and doing even more AI,” Chris said. “Not only are we building and experimenting with all this stuff, but we’re also out hosting events, participating in the community, and sharing our knowledge.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*X6EpMoe6HTjn7PDM" /></figure><h3>What Both of Them Are Carrying</h3><p>Chris describes the job in terms of weight, tracing it back to his time as a 17-year-old guest relations supervisor at Six Flags, where he was hammered into taking care of customers.</p><p>“We’re trusted to deliver on projects that are incredibly important to these businesses. Sometimes the project is their business,” he said. “I think about the employees and their families too. David started this thing 21 years ago, and I want to guarantee it’s here for another 21.”</p><p>David’s confidence rests on the same place it always has. “Slingshot has always changed. That’s why we’re still here 21 years in,” he said. “We’ve had several clients for more than a decade, and the only reason that’s possible is that we change with the times. If we only did what we did 15 years ago, we wouldn’t be here. The important thing to remember is that tech is just a tool; we’re just using that tool to solve business problems.”</p><h3>The Part That Doesn’t Change</h3><p>So here’s the shape of it. A founder who built a company around passionate people and then changed everything else around them multiple times. A reinvention that nearly broke the company and became its destiny. A resident who arrived during the hard part, stayed a decade, and now owns the direction. An AI shift that touched every role, not just a few. And a set of new offerings built on the same old contract: help clients use technology better.</p><p>Ask either of them what Slingshot still is, and the answers land in the same place. “Slingshot is still the kind of company people want to work for, and that companies want to hire for their toughest challenges,” David said.</p><p>“We’ll always be the kind of business you can trust with your idea,” Chris said. “A true partner to your company.”</p><p>Twenty-one years of change produced a company that’s good at changing. The only real question left is what we decide to become next.</p><h3>Inside the LaunchAI Bet</h3><p><a href="https://www.yslingshot.com/introducing-launch-ai/">Read The LaunchAI Story ⇢</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/851/0*Zy04JuVZcLFTEZgc" /></figure><h3>Written by: Savannah Cherry</h3><p>Savannah leads marketing and new business at Slingshot. She writes, posts, and creates all things Slingshot, and helps companies navigate working with a tech partner for the first time. While she isn’t developing software, her CIS minor and a tendency to tinker with AI tools to streamline her own work keep her up to speed on the team’s work. She co-organizes and hosts the Louisville AI Exchange, and she can’t rest until all her work is done.</p><p><a href="https://www.yslingshot.com/author/savannah/">View All Posts</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/851/0*cGvB5RjlMWxHUodN" /></figure><h3>Expert: David Galownia</h3><p>David excels at propelling Slingshot towards their goals and oversees the strategic direction of the company. He’s been described as ‘intense, driven, caring, and passionate’ both at work and play. At work, he enjoys watching his team explore, imagine, and reinvent to do the best by their clients. At play, he drives Karts at insanely high speeds and scares his wife half to death. It’s all or nothing. Which means he gives it all.</p><p><a href="https://www.linkedin.com/in/david-galownia-9202211/">Linkedin</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/851/0*H1ft_XtzEXcaUxsY" /></figure><h3>Expert: Chris Howard</h3><p>Chris has been in the technology space for over 20 years, including being Slingshot’s CIO since 2017. He specializes in lean UX design, technology leadership, and new tech with a focus on AI. He’s currently involved in several AI-focused projects within Slingshot.</p><p><a href="https://www.linkedin.com/in/chris-howard-11084b9/">Linkedin</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8e79adfbf6db" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[6 Single Points of Failure Your Product Cannot Survive]]></title>
            <link>https://medium.com/@heyslingshot/6-single-points-of-failure-your-product-cannot-survive-7683680de65b?source=rss-cc8458c4630f------2</link>
            <guid isPermaLink="false">https://medium.com/p/7683680de65b</guid>
            <category><![CDATA[risk-management]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Slingshot]]></dc:creator>
            <pubDate>Fri, 23 Oct 2026 07:01:24 GMT</pubDate>
            <atom:updated>2026-11-12T16:18:39.447Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*o9WDG7ZqRSvx1a_b.png" /></figure><p>Here’s an uncomfortable exercise: name every system your business depends on, then name the second person who could keep each one running tomorrow morning if something happened.</p><p>For most products, that second name doesn’t exist. The code, credentials, deployment steps, run cost, and the reasoning behind decisions nobody wrote down all sit with one person, and none of it surfaces until that person is unreachable.</p><p>AI didn’t create that risk. It made products cheap enough to start that far more companies now carry it without ever deciding to.</p><h3>Summary</h3><p><em>Locked credentials and undocumented deployments cost time and money, but lost data, a missing codebase, or a security breach can end a product outright. AI can map what an inherited system does, yet only written documentation explains why it was built that way. The fixes cost far less than the exposure: required source control, automated deployments, a second reviewer, backups stored apart from production, a run cost known at 10x volume, and a demo every week or two.</em></p><h3>One Person, One System, No Backup</h3><p>Small teams are no longer a small-company story. They show up in large organizations too, wherever one person owns one system.</p><p>“With AI, we’re going to start having smaller and smaller teams,” said Doug Compton, Director of AI Engineering at Slingshot. “Even in a large company, you can end up with one-person teams or one-person projects. That’s the dilemma.”</p><p>The quality trade-off is structural. On a normal team, a second developer reviews the work before it ships, and that review catches what the author cannot see.</p><p>“That single developer has to be a jack of all trades, and they’re probably a master of none,” said Chris Howard, President of Slingshot. “To create a secure, growth-capable solution, you need an architect. You need a few different mindsets and points of view to make sure that what’s coming out the door is ready for prime time.”</p><p>Then there’s the simplest version of the problem: one developer takes a vacation, loses internet for a few hours, or takes another job. Your product is down, and nobody else can reach the steering wheel.</p><h3>What Walks Out the Door With the Developer</h3><p>Ask most executives where their product dependency lives, and they will say the code. That’s only the first item on the list.</p><p>Credentials come first in practice. If one person holds the login required to deploy, every release waits on that person being reachable and cooperative. Recoverable, usually. Expensive and slow, always.</p><p>The second one is worse. “The one dependency that scares me the most is if the source code only lives on that person’s computer. They weren’t using GitHub, and then they left,” Doug said. “Now the company has zero source code. Depending on the language, you might be able to reverse engineer it, but if it’s compiled, you could be missing it altogether.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*qKzDdlpn4lst9_78" /></figure><p>Deployment is the quiet third. Plenty of systems require several machines to talk to each other in a specific order, with specific configuration. “AI is great at running on your computer and looking at the code on your computer,” Doug said. “But once a deployment relies on multiple machines, the AI probably doesn’t (and probably shouldn’t) have access to those other machines. If your deployment is more complicated than copying files up, that becomes a real problem after the person leaves. You could even break production trying to deploy it.”</p><h3>The Run Cost Nobody Else Can See</h3><p>Here’s a question worth asking before your next board meeting: what does this product cost to run every month, and is there only one person who can tell you?</p><p>“The cost to run your product is something the developer knows and other people may not,” Chris said. “In the context of AI, that dev may have some idea of the number of tokens required and how verbose the whole process is. The business owner won’t realize those things until the product is live; they could be pricing for something completely incorrect.”</p><p>Read that as a pricing problem, not a technical one. If the person who understands your unit economics leaves before you do the math yourself, you find out your margin is wrong after customers are already on the plan. In AI-heavy products, where per-transaction cost scales with usage instead of staying fixed, that gap gets expensive fast.</p><p>The fix costs you one conversation: ask what the system costs to run at current volume, then at ten times current volume. Write the answer down somewhere that isn’t a person’s memory.</p><h3>Backups and Infrastructure Fail Quietly, Then Permanently</h3><p>Most single points of failure give you time to recover. But a couple together doesn’t.</p><p>“There are instances where AI has taken down production,” Doug said. “In one case I saw, the production server was misconfigured; the production database sat on the same hard drive as the backups. Someone erased the drive, losing all production data and all backups. If your lone developer is also in charge of setting up infrastructure and that isn’t their strong suit, you could lose all of your data.”</p><p>That gap between inconvenient and unrecoverable is the one worth understanding before you need it. “Losing your backups could kill the product outright,” Doug said. “If the data is gone, you’ve lost more than the data. Your customers lose faith in you, and there’s no easy way to earn that back. Losing the source code sits right behind it, because recreating a system from scratch takes months, sometimes years.”</p><p>Chris landed somewhere totally different. “The one that would end it for me is security. If you launch with a vulnerability and get hacked, that damages the product and could hurt your whole company. On top of fixing the vulnerability, you have to rebuild trust in your app. You don’t always get to recover from that one.”</p><p>Compare that against the rest of what can go wrong. Locked credentials, undocumented deployment steps, and decisions nobody wrote down all cost you time and money, and time and money are survivable. Lost data, a missing codebase, and a public breach are a different category entirely.</p><h3>AI Fills Part of the Gap, and Widens Another</h3><p>AI genuinely helps here, which is why this problem is easy to underrate. Inheriting an unfamiliar codebase used to be a slow archaeology project. Now it’s a research task you can run in an afternoon.</p><p>“Before AI, you just spent a lot of time reading the code and working to understand it,” Doug said. “With AI in the picture, it can do that research for you. It creates documentation, sequence diagrams, flow charts of how the system works. It makes it a whole lot easier.”</p><p>Access changes too. “If the business owner has access to the code, they can lean on AI to ask questions that in the past would have had to go through the developer,” Chris said. “Used correctly, AI makes that better.”</p><p>Here is the catch: AI can tell you what the system does, but it can’t tell you why. “AI only has a limited memory of what happened in the past, unless it’s thoroughly documented somewhere,” Doug said. “It might re-implement the same thing a customer wasn’t happy about, or repeat a decision that shouldn’t be replicated.”</p><p>Chris put the same gap in business terms. “Unless you have documentation to go along with the code, you never really know why certain decisions were made,” he said. “You don’t know what’s important and what’s not; you only know what was built. You don’t know if it hit the mark as far as what the business needed.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*TE35ZxgVO8-W692c" /></figure><p>That is the part AI inherits rather than solves. Your institutional memory is only as good as what somebody bothered to write down.</p><h3>The Red Flags, and the Cheap Fixes</h3><p>Single points of failure rarely announce themselves, but they do leave traces. Red flags are out there if you know what to look for.</p><p>Start with who built it. If the person maintaining your product moved from the business side into development, they may never have been taught source control, backups, or deployment hygiene. Hence, the gaps stay invisible to them, too.</p><p>Then check what you can state about it yourself, starting with the monthly run cost. And finally, the simplest test of all: sit down and click through the product start to finish. If you have never watched it work end to end, you don’t actually know what state it’s in.</p><p>The good news is that fixing those red flags is cheaper than the exposure. “Have standards,” Doug said. “Identify a source control provider and require that it gets used. Set up continuous integration and deployment so releases are automated and documented. Even on a single-developer project, pull in another developer to review code to make sure things are done properly. And consult experts on infrastructure and backups.”</p><p>“Anyone can vibe code an application and deploy it to Vercel or another host, and IT has no idea it exists,” Doug said. “A non-developer in a department builds something, the team relies on it heavily, and nobody has oversight. That’s a problem waiting to happen.” His fix: keep a log of all the software your company is using and has deployed.</p><p>Chris pushed on visibility and containment. “Have a demo every week or two, so you stay close to what’s being built.”</p><h3>The Risk You Accept Should Be the Risk You Chose</h3><p>Nothing in this conversation required a bigger team. It required knowing where your product actually depends on one person, and deciding on purpose which of those dependencies you are willing to live with.</p><p>Source control, automated deployments, an outside reviewer, a real backup, a known run cost, and a standing demo. Every item is cheap compared to rebuilding a codebase from scratch or explaining lost customer data.</p><p>AI didn’t create this problem. It made one-person products so easy to start that far more companies now carry the risk without ever noticing they picked it up.</p><p>The question isn’t whether you have a single point of failure. It’s whether you can name it before it names itself.</p><h3>Fluency Includes Knowing Where to Stop</h3><p><a href="https://www.yslingshot.com/ai-fluent-team/">AI-Fluent Team Vs An AI-Licensed One ⇢</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/851/0*M_QvMnQT459qoOML" /></figure><h3>Written by: Savannah Cherry</h3><p>Savannah leads marketing and new business at Slingshot. She writes, posts, and creates all things Slingshot, and helps companies navigate working with a tech partner for the first time. While she isn’t developing software, her CIS minor and a tendency to tinker with AI tools to streamline her own work keep her up to speed on the team’s work. She co-organizes and hosts the Louisville AI Exchange, and she can’t rest until all her work is done.</p><p><a href="https://www.yslingshot.com/author/savannah/">View All Posts</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/851/0*-XMwrKaWLxFNB1B8" /></figure><h3>Expert: Chris Howard</h3><p>Chris has been in the technology space for over 20 years, including being Slingshot’s CIO since 2017. He specializes in lean UX design, technology leadership, and new tech with a focus on AI. He’s currently involved in several AI-focused projects within Slingshot.</p><p><a href="https://www.linkedin.com/in/chris-howard-11084b9/">Linkedin</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/851/0*O0u-pOy-VYeP_CaT" /></figure><h3>Expert: Doug Compton</h3><p>Born and raised in Louisville, Doug’s interest in technology started at 11 when he began writing computer games. What began as a hobby turned into his career. With broad interests that range anywhere from snorkeling, science, WWII history and real estate, Doug uses his “down time” to create new technologies for mobile and web applications.</p><p><a href="https://www.linkedin.com/in/doug-compton-1564438/">Linkedin</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7683680de65b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Your Code Was Not Written by AI. It Will Be Attacked by It.]]></title>
            <link>https://medium.com/@heyslingshot/your-code-was-not-written-by-ai-it-will-be-attacked-by-it-ab7e233a6bef?source=rss-cc8458c4630f------2</link>
            <guid isPermaLink="false">https://medium.com/p/ab7e233a6bef</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[supply-chain-security]]></category>
            <dc:creator><![CDATA[Slingshot]]></dc:creator>
            <pubDate>Fri, 16 Oct 2026 07:00:48 GMT</pubDate>
            <atom:updated>2026-11-12T16:17:49.492Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*wDfg4AkwtSfJFExm.png" /></figure><p>For years, many companies have operated on a quiet assumption: nobody is coming for us. Someone else is the bigger target, the more obvious payday, or the easier way in. An attacker would have to spend weeks of skilled, manual effort to get into our systems, and we’re not worth the trouble.</p><p>That assumption held up because it was a math problem. Effort on one side, payoff on the other. When the effort ran high enough, the shortcuts you took stayed survivable.</p><p>AI collapsed one side of that equation. The work that used to take weeks now takes a prompt, which means the calculation that kept you off the list no longer runs in your favor. Your code never changed; the price of attacking it did.</p><p>That shift lands hardest on software written long before any of this existed, built by teams who made reasonable decisions under a threat model that has since expired. Here is what changes for those teams, and the controls worth putting in place first.</p><h3>Summary</h3><p><em>The controls that hold up now are unglamorous: keep packages current, wait a few days before adopting a new release, and vet dependencies as an ongoing discipline rather than a one-time clearance. Enforcement belongs in the CI/CD pipeline, where a failed build forces a decision that an ignored alert never will, with automated AI code review running ahead of human review to catch chained vulnerabilities across files. Point that same analysis at your own codebase on a schedule, because attackers already are.</em></p><h3>The Barrier to Entry Fell, and Your Risk Profile Moved With It</h3><p>Attackers run the same cost-benefit analysis defenders do. Both weigh time and skill against payoff, and for a long time that calculation protected many companies simply by making them inconvenient. Inconvenience is not a defense anymore.</p><p>“The barrier to entry is so low now,” said Steve Anderson, Principal Developer and AWS Architect at Slingshot. “A lot of people took shortcuts before because they thought ‘we’re just a mid-size company, nobody’s really going to target us.’ The targets that get picked now could be just about anybody.”</p><p>Steve pointed to the clearest example: “Writing specific tooling geared toward a particular application used to take a lot of time,” he said. “If you had to handwrite that stuff, it would take forever, and the juice wouldn’t be worth the squeeze. But AI’s lowered the bar on the effort level it takes, so those lower value targets are now something you might go after.”</p><p>Doug Compton, Slingshot’s Director of AI Engineering, framed the escalation in familiar terms. “Before, you had script kiddies: people who weren’t true hackers but found scripts out on the internet they could run,” he said. “Now with AI, that’s on steroids. You can prompt AI to go try something quickly and across multiple sites simultaneously.”</p><p>The uncomfortable part is what that does to your threat model. You didn’t get more valuable; the cost of coming after you simply collapsed.</p><h3>Your Dependencies Became the Front Door</h3><p>Ask most teams where their security risk lives and they’ll point at their own code. The recent wave of supply chain attacks says otherwise.</p><p>“Hackers are putting exploits into third-party libraries,” Doug said. “People update to the latest version and pull in the code that has the security issue in it. Their third-party suppliers end up causing the vulnerabilities in their software.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/633/0*-qIyu3UvvvLXSYZM" /></figure><p>These libraries are attractive targets for a reason, and it has nothing to do with code quality. Many are maintained by volunteers doing the work as a hobby, which means the people guarding a dependency thousands of companies rely on are stretched thin, underappreciated, and operating without a security team.</p><p>That exposure compounds with time. Developers have always let libraries and third-party tools drift a few versions behind, and that drift used to be a tolerable trade. Now it’s the opening.</p><p>Which is where there’s some good news for teams running older systems. The upgrade work that used to be impossible to justify has gotten cheap. “Updating to the latest version of a particular framework tends to actually get approved now, because it doesn’t take as long,” Steve said. “They’re pretty quick turnarounds.”</p><p>Doug pairs that speed with a simple, almost boring defense: stop treating ‘latest’ as a synonym for ‘safe.’ “One rule of thumb is to wait at least three days to a week before you upgrade,” he said. “Let the new version sit out there for a bit, so if something nefarious happened to it, there’s an opportunity for it to get announced or fixed by then.”</p><p>Steve added two more open doors teams rarely consider. The first is letting npm install run binaries, which lets a package you haven’t inspected execute arbitrary code. The second is container images. “If you pull a malicious Docker image and use it in a CI build, that’s also a supply chain attack,” he said. Roll that same tainted image into production, and you have handed someone a path to data exfiltration.</p><h3>AI Reads Code the Way Attackers Wish They Could</h3><p>Single-vulnerability exploits are the easy case, since scanners can catch the known ones. The harder case is when three or four unremarkable things line up across a codebase.</p><p>“AI is much better at analyzing multi-chained exploits, where it’s not just one vulnerability,” Doug said. “It might be three or four different things working together. Those can span several files, making it hard for a human to remember all the pieces at once. AI puts those pieces together a little more easily.”</p><p>Read that again from a defender’s perspective, because it cuts both ways: If an attacker can run a model across your exposed source and surface chained vulnerabilities in an afternoon, you can do the same thing first. The advantage goes to whoever looks harder, and right now, attackers are the ones looking.</p><h3>The Social Layer Is Part of the Attack Surface Now</h3><p>The piece of this that has nothing to do with your infrastructure is the one most likely to catch you off guard.</p><p>Doug described a case involving a widely depended-on open source library maintained by one person in his spare time. “He noticed a pull request that was very suspicious, and when he decided not to pursue it, he got about a dozen emails from other ‘people’ pushing him to approve it,” he said. “After they dug in further, they found out all those messages were AI. AI submitted a pull request with an exploit, and when it got turned down, it spawned other accounts that emailed the maintainer, pressuring him to accept it.”</p><p>An exploit, a rejection, and a coordinated pressure campaign, all automated and aimed at one tired volunteer.</p><p>The defenses built to stop that kind of volume are weakening too. “AI’s ability to solve CAPTCHAs now is pretty good,” Steve said. “That just accelerates automated account creation, or makes it easier to accomplish.” The guardrail that assumed bots couldn’t read a distorted word is holding less every month.</p><h3>Vetting Packages Is a Judgment Call, Not a Checkbox</h3><p>So how do you decide whether a package deserves a place in your build?</p><p>Start with the crowd. “I’d shy away from smaller libraries and stick with ones that have large support groups and lots of users,” Steve said. “Lots of eyes, lots of people doing code reviews. Something malicious is less likely to sneak in, and the quality in general should be better than a solo dev working on it on weekend nights.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/630/0*wUIpCMcIewM-4L6x" /></figure><p>That heuristic holds most of the time, but it has a blind spot worth naming. Download counts measure popularity, not oversight. A package can sit under thousands of production builds and still come down to one person reviewing pull requests on a Tuesday night. Widely used and widely watched aren’t the same thing.</p><p>Doug pushes further for anything business-critical. “Depending on how critical it is, doing full code reviews on the software you use is one way to do it, assuming the code is available,” he said. “But so many of these projects use thousands upon thousands of different libraries, so it’s hard to do that. If it’s critical for your company, create a database of which ones have been reviewed and found clean.”</p><p>The ceiling is real. Nobody reviews every dependency in a large project. But the handful sitting on your critical path can earn that scrutiny, and knowing which ones you have actually cleared beats assuming all of them are fine.</p><p>And Doug wants teams to stop misreading a clean scan. “Just because you run an update on your packages and it says there are no critical vulnerabilities, that doesn’t necessarily mean that’s the case,” he said. “It might just mean nothing has been reported yet; one may get discovered afterward.”</p><p>A package marked secure today might not be next week. That’s what makes vetting a discipline, not a decision. You’re not clearing a package once and filing it away; you’re committing to keep looking at something you already approved.</p><h3>Build the Discipline Into the Pipeline, Not the To-Do List</h3><p>Everything above fails the same way. Someone sees the vulnerability notice, decides it’s not today’s problem, and moves on. Doug sees it constantly. The alerts fire, and people ignore them.</p><p>The fix is to stop relying on attention and start relying on the pipeline. “Your CI/CD pipeline can do a lot of that gathering for you,” he said. “If your pipeline sees on a code commit that a package has changed, it can check how old it is and stop that commit from happening.” Same gate for known vulnerabilities. The build fails, someone has to act, and the decision is no longer optional.</p><p>Steve layers automation on top of it. “ <a href="https://github.com/dependabot">Dependabot </a>will actually auto-upgrade those packages for you,” he said. “Get a nice test suite in place, and when the packages upgrade and the tests pass, you have more confidence going forward.”</p><p>Those two instincts pull in opposite directions, so configure the cooldown instead of letting every new release land the day it ships. The tests tell you the upgrade didn’t break anything; they don’t tell you the package is clean.</p><p>Then there is the review step that most teams haven’t added yet. Doug suggests adding an automated AI code review in the deployment pipeline: “It doesn’t replace a human code review,” he said. “But AI is good at finding things a human would overlook, especially when several files interact in a weird way that opens up an exploit. Run the AI code review first, let the developer address those issues, then do the human review after that.”</p><p>And it’s worth extending that same thinking past the commit. A monthly security audit of your existing code, run by AI rather than waiting for a human to find the time, catches what shipped before anyone was watching. Static analysis tools like <a href="https://www.sonarsource.com/products/sonarqube/">SonarQube </a>have covered part of this for years. The point is that something, human or machine, needs to be evaluating vulnerabilities on a schedule. And it’s best if it’s both.</p><h3>What Actually Changed</h3><p>Strip away the new vocabulary and the shift is narrow but sharp: attackers got faster, cheaper, and more persistent. Your software didn’t change, but your assumptions about who would bother coming after it did.</p><p>The single most important control isn’t new. It’s almost disappointingly simple: “It’s the same as it was before AI,” Steve said. “Update your packages. There’s just more of it now.”</p><p>That one control sits alongside everything else here. Audit your dependencies, treat package vetting as ongoing rather than one-time, remember that social pressure is now part of the payload, let the pipeline enforce what attention won’t, and point the same analysis capability at your own code before somebody else does. None of it is exotic, but all of it assumes you act before you have a reason to.</p><p>The advice didn’t change. The consequences of ignoring it did. Patching lagging dependencies has always been the item that slides to the next sprint, the ticket nobody fights for, the work that feels safe to defer because it has been for years. That window closed. The effort to find and exploit what you left sitting there has dropped far enough that someone will get to it before you do.</p><p>That is the uncomfortable truth: the defense hasn’t changed. The deadline has.</p><h3>What Production Actually Buys You</h3><p><a href="https://www.yslingshot.com/ai-prototype-vs-production/">Prototypes Prove It Can Work. Production Proves It Does. ⇢</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/851/0*Wkh5ORwG-NkthBgr" /></figure><h3>Written by: Savannah Cherry</h3><p>Savannah leads marketing and new business at Slingshot. She writes, posts, and creates all things Slingshot, and helps companies navigate working with a tech partner for the first time. While she isn’t developing software, her CIS minor and a tendency to tinker with AI tools to streamline her own work keep her up to speed on the team’s work. She co-organizes and hosts the Louisville AI Exchange, and she can’t rest until all her work is done.</p><p><a href="https://www.yslingshot.com/author/savannah/">View All Posts</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/851/0*OicDIA2Q2nqW6GlP" /></figure><h3>Expert: Doug Compton</h3><p>Born and raised in Louisville, Doug’s interest in technology started at 11 when he began writing computer games. What began as a hobby turned into his career. With broad interests that range anywhere from snorkeling, science, WWII history and real estate, Doug uses his “down time” to create new technologies for mobile and web applications.</p><p><a href="https://www.linkedin.com/in/doug-compton-1564438/">Linkedin</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/851/0*0hYxRFZiWnVEccc3" /></figure><h3>Expert: Steve Anderson</h3><p>Steve is one of our AWS certified solutions architects. Whether it’s coding, testing, deployment, support, infrastructure, or server set-up, he’s always thinking about the cloud as he builds. Steve is extremely adaptable, and can pick up the project and run with it. He’s flexible and able to fill in where needed. In his spare time, he enjoys family time, the outdoors and reading.</p><p><a href="https://www.linkedin.com/in/steve-anderson-5b13bb1/">Linkedin</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ab7e233a6bef" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[3 Shifts That Separate an AI-Fluent Team From an AI-Licensed One]]></title>
            <link>https://medium.com/@heyslingshot/3-shifts-that-separate-an-ai-fluent-team-from-an-ai-licensed-one-ec70a187554f?source=rss-cc8458c4630f------2</link>
            <guid isPermaLink="false">https://medium.com/p/ec70a187554f</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[generative-ai-tools]]></category>
            <category><![CDATA[ai-licensing]]></category>
            <category><![CDATA[ai-tools]]></category>
            <dc:creator><![CDATA[Slingshot]]></dc:creator>
            <pubDate>Tue, 13 Oct 2026 20:56:01 GMT</pubDate>
            <atom:updated>2026-10-13T20:56:01.718Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*us0XaQoGRkTp4yWLsOZQdA.png" /></figure><p>Picture a company that bought a few hundred AI licenses. A year goes by; what would tell you whether any of it mattered?</p><p>Most leaders reach for the usage report. That instinct is right, and it’s also nowhere near the finish line. Seats get filled, logins happen, and somebody in marketing makes a very good meme. None of that proves your team changed how it works.</p><p>Licenses prove access, but fluency proves something harder to buy: your people know which problems belong to AI, which ones belong to your developers, and which ones should never leave the building.</p><p>Here are the three shifts that show up when a team crosses that line, plus the one boundary a fluent team never crosses.</p><h3><em>Summary</em></h3><p><em>Fluent teams show it in three places: the first prototype arrives from the person with the problem instead of the dev team, work that always got cut (accessibility, observability, internal tooling) finally gets built because the economics changed, and people reach for AI on problems nobody assigned them. That last one predicts the other two, because pattern recognition comes from repetition, not from a mandatory workshop. The boundary holds steady around consumer data, public-facing systems, and long-term maintenance, where human expertise still owns the call.</em></p><h3>Start With a Baseline, Not a Vibe</h3><p>Before you can spot any shift, you need something to compare it against. Most companies skip that part and end up arguing about feelings.</p><p>“The first thing I’d look at is the usage report,” said Steve Anderson, Principal Developer and AWS Architect at Slingshot. “Purchasing licenses is one thing. Are people actually using them? And I wouldn’t wait to check.”</p><p>Usage answers the easy question. Chris Howard, Slingshot’s President, tracks what comes after it. “That’s what we did early on. We watched usage, but what follows that is the real question: are you seeing processes change?”</p><p>Process change is the signal. Measuring it requires groundwork that most teams never lay.</p><p>“You have to have baseline measurements, or it’s very difficult to assess impact,” Steve said. “That’s not specific to AI licenses; that’s any change. Without a baseline, I don’t know how you measure improvement.”</p><p>Skip the baseline and your ROI story collapses into anecdote. Productivity <em>feels</em> faster. Quality <em>feels</em> better. But neither claim survives a board meeting.</p><p>One signal shows up even without instrumentation, and it has nothing to do with speed.</p><p>“Most of the time, people who are fluent in AI take on more complex problems,” Chris said. “The AI handles the manual work, so your thinking gets elevated. You can pay more attention to the content of a presentation and how it’s going to land, and less on how you build the presentation.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/629/0*Ke2UX6LMcWk8yiPS" /></figure><p>That’s the tell. A licensed team does the same work slightly faster. A fluent team moves up a level and starts solving things it used to postpone.</p><h3>Shift One: The Prototype Shows Up Before the Project Does</h3><p>The clearest structural change is who builds the first version. It’s no longer the development team; it’s the person who has the problem.</p><p>“It used to be that people would come to us with an idea and we’d prototype it, hand it back, and iterate,” Chris said. “We were the ones adjusting and tweaking. Now the customer comes to us with a concept and a prototype already built.”</p><p>Slingshot’s job shifted accordingly. Instead of translating a description into a first draft, the team applies technology and UX expertise to something that already exists, elevating it even further.</p><p>That reordering pays off immediately, because the expensive part of software has never been the code; it’s been the misunderstanding.</p><p>“It closes the communication divides,” Steve said. “You can click through someone’s prototype and get a real feel for what they were thinking. There’s less iteration, because it’s already sitting there, looking you in the face. Multiple rounds of iteration come at a time and money cost.”</p><p>Inside your own organization, the same shift is already underway. Your operations lead stops filing a requirements doc and starts sending a working click-through. Your CFO stops describing the dashboard and starts showing you one. That changes what your intake process needs to handle.</p><p>Chris sees the next step forming now. “We’re standing up what I’d describe as a fourth environment: a sandbox. The customer can use their own cloud instance to modify that sandbox, try different concepts, and our team can see what they came up with and do something with it.”</p><p>The prototype stops being a picture a developer glances at. It starts informing the entire design, and both sides keep editing it.</p><h3>Shift Two: The Work That Always Got Cut Now Gets Built</h3><p>Every CIO has a mental list of the items that die in every scoping conversation: observability, accessibility, internal tooling. The work everyone agrees matters, and nobody wants to fund.</p><p>That list of what ends up on the cutting room floor is getting shorter. “Ops tools and observability tools are hard to communicate the value of, and they’re usually very quick to be on the chopping block,” Steve said. “Now we’re able to implement them for a fraction of the cost, and teams are now approving that tooling for their applications.”</p><p>The same economics apply to quality work that used to lose the budget fight. “Making a web app fully responsive is much easier than it used to be,” Chris said. “Same with accessibility; that almost always got cut in the past.” Steve added a detail worth sitting with: “LLMs do that without you even asking them to.”</p><p>Read that as a governance opportunity, not a convenience. The standards you’ve been deferring for years now cost a fraction of what they did. If accessibility or observability still isn’t in your builds, the constraint is no longer budget.</p><p>Fluent teams also rebuild the connective tissue between conversations and code. “You can lean on Zoom, Otter, or any transcription service to record your project meetings, then ingest the transcripts into your specs,” Chris said. “You talk with a customer during the sales cycle and hear what’s important to them. Now you can hand all of that knowledge to the project team.”</p><p>Context that used to evaporate between sales and delivery now travels with the project. That’s a process change you can actually point to.</p><h3>Shift Three: Fluent People Use AI When Nobody Asked Them To</h3><p>The third shift never appears in a dashboard, and it predicts the other two better than anything else does.</p><p>Steve talked about a time when he used AI for personal reasons: “My family had seven options from a travel agent, and my kid was hyper-focused on the Bahamas,” Steve said. “I fed the seven options in, told AI to scrub the locations, and asked for a slideshow with pros and cons I could present to her so she could pick.”</p><p>Chris went further and built infrastructure for his own mornings. “I wear an Oura Ring, so I built an MCP for it and plugged it into my Claude instance. Every morning I get a daily brief: my calendar, emails I missed, AI news locally and globally, how I slept, and how ready I am for the day.”</p><p>He put the scale of the change plainly. “At this point it has entirely replaced everything I used to do with Google.”</p><p>Weekend tinkering looks like trivia until you notice what they produce. “Personal experimentation shows you know what a problem AI can solve looks like,” Chris said. “You know where it best fits.” Steve framed it as range: “It shows you’re thinking about applications beyond the obvious, instead of staying inside the box with it.”</p><p>Pattern recognition is the actual skill, and you can’t train it in a mandatory workshop. It comes from repetition across enough contexts that people stop guessing.</p><p>Personal use also delivers the lesson no vendor demo will. “I did some trip planning for Costa Rica, and it gave me prices for hotels,” Chris said. “When I went to book, it had understated the price by about four times. A hotel it said was $400 a night was $1,800. You start to realize you can’t fully trust everything it tells you.”</p><p>Better to learn about hallucinations on a hotel booking than on a client deliverable.</p><h3>Fluency Also Means Knowing Where to Stop</h3><p>Enthusiasm creates its own risk. The more capable the tool feels, the more people want to connect it to everything they own.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/629/0*RQRwUT9jbaPjtbIh" /></figure><p>“Once you get into AI, you think ‘this thing is powerful, I’m going to hook all my things up to it,’ but that’s a problem,” Steve said. “Not having proper guardrails and safety gates in place is a mistake an AI-fluent person shouldn’t be making.”</p><p>Knowing what not to automate is part of the fluency, not a limit on it. “Anything touching consumer data needs proper controls,” Steve said. “Encryption, security, all of it. That isn’t something I’d completely outsource to AI. If you put something into the wild, you’re taking responsibility for it, so you need to understand what you built.”</p><p>Chris draws the line at exposure. “As soon as it’s public-facing, and especially public-facing with sensitive data, you want experts in that space looking it over, critiquing it, and prepping it for production. It doesn’t mean you can’t build it, just that it needs a review process.”</p><p>Two failure modes follow the same internal tools your team is now happily generating. The first arrives quietly. “Nobody really thinks about support and maintenance long-term. They don’t have a plan for it,” Chris said. “With the security vulnerabilities out there, that’s a fast-evolving world you have to keep up with.”</p><p>The second mode arrives on your best day. “You put out your MVP with certain assumptions about scale. If it explodes with popularity, that’s a blessing and a curse,” Steve said. “What you built wasn’t built for that scale of user base. The LLM built it just fine, but it didn’t prepare for a massive influx of people.”</p><p>Just because code compiled doesn’t mean it’s suited for the big leagues. Having a team ready to help can make or break your product over the long term.</p><h3>Fluency Is a Behavior, Not a Purchase</h3><p>Three shifts tell you where your team actually stands. First versions start arriving from the people with the problem instead of the people with the keyboards. Work that always lost the budget fight finally gets built. And your team reaches for AI on problems nobody assigned them, which is how they learn where it fits and where it falls apart.</p><p>Around all three sits the boundary that separates confidence from exposure: consumer data, public-facing systems, long-term maintenance, and anything carrying someone else’s personal information stays under human expertise.</p><p>None of that arrives with the invoice. Usage reports tell you people logged in, baselines tell you whether anything improved, and behavior tells you the truth.</p><p>You can buy every license on the market by Friday. You cannot buy the judgment to know which problems are worth handing over.</p><h3>More on How Fluency Buys Back Time</h3><p><a href="https://www.yslingshot.com/ai-saves-developer-time/"><strong>4 Places AI Buys Back Your Developers’ Time ⇢</strong></a></p><h3>Written by: Savannah Cherry</h3><p>Savannah leads marketing and new business at Slingshot. She writes, posts, and creates all things Slingshot, and helps companies navigate working with a tech partner for the first time. While she isn’t developing software, her CIS minor and a tendency to tinker with AI tools to streamline her own work keep her up to speed on the team’s work. She co-organizes and hosts the Louisville AI Exchange, and she can’t rest until all her work is done.</p><p><a href="https://www.yslingshot.com/author/savannah/"><strong>View All Posts</strong></a></p><h3>Expert: Chris Howard</h3><p>Chris has been in the technology space for over 20 years, including being Slingshot’s CIO since 2017. He specializes in lean UX design, technology leadership, and new tech with a focus on AI. He’s currently involved in several AI-focused projects within Slingshot.</p><p><a href="https://www.linkedin.com/in/chris-howard-11084b9/"><strong>Linkedin</strong></a></p><h3>Expert: Steve Anderson</h3><p>Steve is one of our AWS certified solutions architects. Whether it’s coding, testing, deployment, support, infrastructure, or server set-up, he’s always thinking about the cloud as he builds. Steve is extremely adaptable, and can pick up the project and run with it. He’s flexible and able to fill in where needed. In his spare time, he enjoys family time, the outdoors and reading.</p><p><a href="https://www.linkedin.com/in/steve-anderson-5b13bb1/"><strong>Linkedin</strong></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ec70a187554f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Fastest Way to Learn AI Is to Build Something With It]]></title>
            <link>https://medium.com/@heyslingshot/the-fastest-way-to-learn-ai-is-to-build-something-with-it-b0d08fd81dcd?source=rss-cc8458c4630f------2</link>
            <guid isPermaLink="false">https://medium.com/p/b0d08fd81dcd</guid>
            <category><![CDATA[hands-on-tutorials]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[education]]></category>
            <category><![CDATA[conference]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Slingshot]]></dc:creator>
            <pubDate>Mon, 12 Oct 2026 20:56:01 GMT</pubDate>
            <atom:updated>2026-10-12T20:56:01.782Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*B9p0BGcWze9-h86xm7jE8w.png" /></figure><p>Ask ten people how to start learning AI, and you will get ten different directions to go. Newsletters, podcasts, a webinar someone swears by, a course that guarantees it’ll all click. Plenty of that is genuinely useful.</p><p>The trouble shows up the second you sit down to use it. You can explain what a context window is and still have no idea what to do when your agent returns confident nonsense on a Thursday afternoon. That kind of knowledge doesn’t transfer through a slide. You build your way into it, or you don’t get it at all.</p><p>Our team spends a lot of time in this gap, both in client work and in the rooms where people come to learn. A few patterns keep surfacing.</p><h3><em>Summary</em></h3><p><em>AI skill comes from pointing a real problem at a tool and working through what breaks, not from collecting vocabulary in a lecture hall. The durable part is the approach: how you frame a problem, how much context you give, and how you recover when the output goes sideways. Tooling will expire, but judgment built in public with other people trying the same thing is what carries into the next model release.</em></p><h3>Start With a Problem You Already Have</h3><p>“Find a problem you’re looking to solve,” said Sarah Bhatia, Director of AI Product Innovation at Slingshot. “Look for a pain point in your day or week, and then start exploring whether something in the AI space could solve it.”</p><p>A real problem gives you two things a tutorial never will: a finish line and a feedback loop. You know it works because the annoyance goes away. You know when it doesn’t because you’re still doing the annoying task by hand.</p><p>That framing also protects you from the biggest beginner trap: treating AI as a subject to study rather than a tool to point at something. Studying AI produces opinions. Pointing it at your Tuesday morning report produces skill.</p><p>Stef Hogan, Product Lead at Slingshot, builds on that with a second move. “Start getting your hands dirty and actually using it, but also find a group of people you’re comfortable with to talk through the hard parts and the confusing parts,” she said. “There’s a lot of power in building in public and building with community.”</p><h3>Nobody Has This Figured Out, and That Helps</h3><p>Software has always carried a myth about the solo operator. “But we’re facing a unique landscape now, on the front line of this technology, with the most pressure around using it,” Sarah said. “It’s been really important to find spaces where we can not only commiserate around the frustration, but find solutions together.”</p><p>The pace is what breaks the old model. Tech workers adapt to change for a living, but nobody has watched a technology shift this broadly this fast. Expertise decays in months. The person who solved your exact problem last week is more valuable than the documentation.</p><p>Stef named the takeaway she keeps leaving AI events with. “No one has this figured out,” she said. “At conferences about other areas of tech, people do have it figured out: Agile, project management, you walk away with something to work with. With AI, it’s the validation that it’s moving quickly, that it’s okay you don’t have it figured out. And here are some tools to keep problem-solving faster.”</p><p>That is not a consolation prize. Knowing the whole field is improvising frees you to improvise.</p><h3>The Mindset Travels, the Tooling Expires</h3><p>New agents ship weekly. Model capabilities jump in ways that make last quarter’s workflow look quaint. So people reasonably assume that learning AI means chasing the stack.</p><p>“We get really hung up on the tooling,” Stef said. “My setup is completely different than Sarah’s. You talk to someone else and think, ‘shoot, my setup is different, I should switch.’ You don’t have to do the same thing as another person. It’s more about the first principles of how you approach what you’re doing.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/629/0*-pbHNbXiPw5h059z" /></figure><p>The durable part sits underneath the tools. “Approaching a problem a certain way, breaking it down a certain way, the importance of context, the importance of understanding how to troubleshoot when things go wrong,” Stef said. “That stays the same.”</p><p>Tooling matters eventually. Once you know what you’re doing, preferences get sharp and specific, and the right tool for a task saves real hours. But you can’t develop that judgment by reading comparison charts; you develop it by breaking something and fixing it.</p><h3>Adults Already Learn by Doing</h3><p>There is a strange inversion in how we teach grown people.</p><p>“When we were in K-12, you’re mostly talked to the whole time,” Stef said. “You’re lectured, then you go do the worksheet. As adults, it’s completely flipped. You’re in a job, and it’s, ‘Here’s this thing, do it.’ It rewires our brain to learn by doing. So I don’t understand why conferences for adults aren’t do-and-learn.”</p><p>Every professional in the room has spent a decade or more learning by getting handed something unfamiliar and figuring it out. Then we put those same people in chairs for eight hours and ask them to retain material they won’t touch until next week, if at all.</p><p>Sarah pointed at what makes the difference between a good day and a wasted one. “Conferences can be good for being exposed to new technology and new solutions, but I want to know that going into it,” she said. “If I show up expecting to really learn something, I expect to walk away with something tangible.”</p><p>The fix is not complicated. Put the tool in their hands while the expert stands there. That reasoning shaped the AI workshop our team is co-hosting in Louisville this October: <a href="https://www.eventbrite.com/e/proof-ai-workshop-ai-in-practice-not-just-theory-tickets-1995766295715">PROOF AI</a>. Whitney Powell, Marketing Manager at Slingshot, helped build the format by working backward from the days that didn’t work. “Going to these conferences, it’s a lot of back-to-back packed lectures,” she said. “It’s hard to keep people’s attention throughout the day, and packing all that in can be very overwhelming. That was the first thing we decided to throw out.”</p><p>What replaced it is a day built around working sessions, where each track hands attendees something to build, and the people who build for a living stay in the room while they do it.</p><h3>Watching Someone Troubleshoot Beats Watching Them Succeed</h3><p>Polished AI demos teach almost nothing, because the interesting decisions already happened offstage.</p><p>Stef is deliberately doing the opposite in her <a href="https://www.eventbrite.com/e/proof-ai-workshop-ai-in-practice-not-just-theory-tickets-1995766295715">PROOF AI</a> session. “It’s going to be very real life, meaning it’s not going to work,” she said. “There’s going to be things I have to troubleshoot in front of people, and that’s the point of the topic: for people to see the way I approach the problems, and how it might be more advanced or different from how they would approach it, but how it gets to the outcome.”</p><p>That is the part nobody writes down: the recovery. The moment you realize the model doesn’t have the context that you assumed it had, and you decide what to change.</p><p>Sarah is running at the same idea from the strategy side, and she thinks most people are still stuck in the first gear of AI use. “Up to this point, most people have engaged with AI from an efficiency standpoint, trying to speed up their work and do more in less time,” she said. “Speed is the least interesting way to use these tools. Phase two is about quality. How do we use these tools to go deeper, to create stronger output, by capitalizing on our unique human insight?”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/629/0*MIo4anf4QERajyOO" /></figure><h3>You Should Leave With Something That Works</h3><p>The best measure of a learning day is what still exists the next morning.</p><p>Whitney is building her session around exactly that. “Somebody is going to bring a specific task they’re dealing with weekly or daily, and we’ll help them craft it into a working skill in Claude, ChatGPT, Copilot, or whatever tool they use,” she said. “Then we give them the framework to go do it again with another repeatable task.”</p><p>The same logic shaped how the <a href="https://www.eventbrite.com/e/proof-ai-workshop-ai-in-practice-not-just-theory-tickets-1995766295715">PROOF AI</a> Workshop ends. Instead of a final keynote, the day finishes with attendees showing each other what they made, plus a wall where people post what they built as they go. “These people are going to be there all day building stuff they should be proud of,” Whitney said. “So everybody gets the opportunity to share what they were able to build. It gives people insight into how they could go back and do that with something else at work.”</p><p>Sarah described where the <a href="https://www.eventbrite.com/e/proof-ai-workshop-ai-in-practice-not-just-theory-tickets-1995766295715">PROOF AI</a> format came from, and the analogy lands. “It came about organically from a bunch of people who’ve gone to AI conferences and wanted to put one together that we’d actually enjoy,” she said. “It’s like being the last of your friends to get married, so you put together a wedding with all the stuff you liked and leave out the stuff you didn’t.”</p><h3>What Actually Sticks</h3><p>The through line is friction. Not the bad kind that wastes your afternoon, but the productive kind that comes from choosing a real problem, getting stuck on it in front of other people, and working out what to do next.</p><p>Reading about AI gives you vocabulary. Building with it gives you judgment, and judgment is the only thing that survives the next model release. Your problem is the curriculum. The people around you are the support system. The thing you get working by the end of the day is the proof.</p><p>So the question is not whether you understand AI yet; it’s what you’ll make it do before next Friday.</p><h3>PROOF AI tickets are available now!</h3><p><a href="https://www.eventbrite.com/e/proof-ai-workshop-ai-in-practice-not-just-theory-tickets-1995766295715"><strong>Register Here ⇢</strong></a></p><h3>Written by: Savannah Cherry</h3><p>Savannah leads marketing and new business at Slingshot. She writes, posts, and creates all things Slingshot, and helps companies navigate working with a tech partner for the first time. While she isn’t developing software, her CIS minor and a tendency to tinker with AI tools to streamline her own work keep her up to speed on the team’s work. She co-organizes and hosts the Louisville AI Exchange, and she can’t rest until all her work is done.</p><p><a href="https://www.yslingshot.com/author/savannah/"><strong>View All Posts</strong></a></p><h3>Expert: Sarah Bhatia</h3><p>Sarah Bhatia brings people together. In her decade plus of product and product-adjacent experience, her focus has been on cross-functional collaboration, asking lots of questions, and getting big results. She excels at strategy development, and getting the right brains in the room to solve big problems. Sarah would describe herself as a daredevil, because she’s not afraid to ask “dumb“ questions, get smart answers, and take (calculated) risks.</p><p><a href="https://www.linkedin.com/in/sarahbhatia/"><strong>Linkedin</strong></a></p><h3>Expert: Stef Hogan</h3><p>Stef Hogan is a product and engineering leader at Slingshot, focusing on product, technology, and AI. Previously, she was the Director of Engineering and Product Operations at a payments company, leading initiatives on shopper experiences, authentication, and cloud migration. She is recognized for her expertise in AI enablement and is active in Louisville’s AI and builder community, fostering experimentation with emerging technologies.</p><p><a href="https://www.linkedin.com/in/stefanie-hogan/"><strong>Linkedin</strong></a></p><h3>Expert: Whitney Powell</h3><p>Whitney earned her degree in Marketing and Management from the University of Kentucky and discovered her passion for marketing and events. Her go-getter attitude, willingness to learn, and problem-solving abilities elevate the Slingshot team. Known as a daredevil, Whitney loves trying new things and embracing challenges, whether traveling to new places or taking on new projects at work.</p><p><a href="https://www.linkedin.com/in/whitney-powell-/"><strong>Linkedin</strong></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b0d08fd81dcd" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Prototypes Prove It Can Work. Production Proves It Does.]]></title>
            <link>https://medium.com/@heyslingshot/prototypes-prove-it-can-work-production-proves-it-does-d4939339a932?source=rss-cc8458c4630f------2</link>
            <guid isPermaLink="false">https://medium.com/p/d4939339a932</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[prototyping]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Slingshot]]></dc:creator>
            <pubDate>Sun, 11 Oct 2026 20:46:02 GMT</pubDate>
            <atom:updated>2026-10-11T20:46:02.054Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xDZuy_pWKKaj9DmWObqfkw.png" /></figure><p>Someone on your team clicked through a demo last week. Screens moved. Data landed where it belonged. The flow made sense on the first try. And then the reasonable question came up, the one every executive asks: if it already works, why does the real version take another X months?</p><p>That question deserves a real answer, because the demo wasn’t a trick. It worked. It just answered a different question than the one production answers. A prototype proves an idea can work. Production proves it does it every day with real users and messy data.</p><p>Here’s what actually happens between those two things, and why the gap keeps surprising smart teams.</p><h3><em>Summary</em></h3><p><em>Today’s prototype is no longer a disposable visual. It functions as a source artifact that carries much of the specification inside it, which is why Slingshot’s teams now field roughly a tenth of the clarification questions developers used to ask. The real spend goes toward what never appears on screen: the guardrails that make an AI answer consistent the thousandth time, the error handling for every path that isn’t the happy one, and the expert judgment to strip out features that were easy to generate but wrong to keep.</em></p><h3>The Demo Works, and That’s Exactly the Problem</h3><p>Prototypes no longer look like sketches and wireframes. They look like shipped software, and clients respond to them accordingly.</p><p>A recent engagement made that clear. The client couldn’t pinpoint where users were dropping off during onboarding, so the team built a quick prototype that showed every recommendation working together as one flow.</p><p>“They initially thought it was a completed product,” said Rachel Foster, Principal Product Designer at Slingshot. “They’d never seen such an advanced prototype, so it was a big shift for their conceptual mind.”</p><p>Some of them assumed it was already running inside their own system. It wasn’t connected to anything. Nothing was wired up. The whole thing mimicked a working product convincingly enough that the distinction stopped registering.</p><p>That reaction isn’t a failure of intelligence. It’s a failure of the old signals. A static Figma layout announced its own incompleteness. Nobody mistook a gray rectangle for a shipped feature. Today’s prototypes removed that signal entirely, and they did so just as clients gained the ability to build convincing prototypes themselves.</p><h3>Production Buys You Everything You Can’t See</h3><p>None of the work that turns a demo into a real product shows up on screen. That’s exactly why it reads like padding on an invoice.</p><p>“We’ve been calling a prototype the tip of the iceberg,” said Sarah Bhatia, Director of AI Product Innovation. “What you see and feel is a very real experience. But you’re looking at a single happy path, and it’s missing everything underneath that makes something production quality.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/633/0*EpYjyNYNqfKs9Rru" /></figure><p>Underneath is where the software handles real data that shows up messy and at volume. It’s where authentication decides who sees what. It’s where the system catches errors and absorbs whatever strange thing a user just did. It’s where monitoring and logging tell you what broke at 2 a.m.</p><p>Chris Howard, Slingshot’s President, points to the piece that trips up AI features specifically. “Clients overlook the guardrails,” he said. “Getting a consistent answer every single time takes a lot more work.”</p><p>For an AI product, consistency is the product. One brilliant answer in a demo doesn’t tell you anything about the thousandth. Customers won’t grade you on the best output they ever saw. They’ll grade you on the worst one.</p><h3>The Real Build Starts With Everything the Demo Skipped</h3><p>Day one of production doesn’t start with code. It starts with writing down every rule the prototype never had to follow.</p><p>“When we finish a prototype, we write extensive PRDs and feature specs,” Rachel said. “One I wrote recently ran over 57 pages. It includes all logic, edge cases, and every detail needed to inform exactly how something should function. That level of instruction is invisible in a prototype.”</p><p>Prototypes show the path a user takes when everything goes right. Production has to define what happens on every other path, and those paths far outnumber the happy one.</p><p>“There are a lot of unseen rules for how software works,” Sarah said. “When we’ve done a good job, the software feels simple and straightforward to the user. That’s part of the disconnect.”</p><p>Good software hides its own complexity. That’s the whole craft. It also means the products your team uses every day have trained everyone to underestimate what sits behind a clean interface. The better the finished software you’ve used, the less you can see of what it took.</p><h3>Your Prototype Is Now the Fastest Spec You’ll Ever Write</h3><p>Here’s the part that changes the math in your favor, and it’s the reason this shift is worth the confusion.</p><p>“A prototype used to be a visual you were validating,” Chris said. “Now it’s a source artifact. It’s documentation that accelerates the build.”</p><p>That’s a real change in what a prototype is worth. It used to get admired, approved, and then broken back down into written requirements before anyone could build from it. Now the prototype carries a large share of the specification inside itself.</p><p>Sarah’s teams spend more time upfront aligning on requirements and documenting them than ever, and the payoff shows up downstream. Rachel put a number on it: the clarification questions she fields from developers have dropped to roughly 10% of what they used to be. Sarah landed on the same estimate independently.</p><p>The effort split still favors production, and it should. Chris puts it at about three units of effort to reach a working prototype and seven to reach something people depend on. That ratio isn’t waste. It’s the reason the seven goes faster than it used to.</p><h3>Easy to Build Is Not a Reason to Build</h3><p>Cheap features create a new problem. When adding something takes an afternoon, “let’s add it” becomes the default answer to every gap.</p><p>“It’s easier than ever for the answer to a problem to be, we’ll just add another feature,” Sarah said. “Part of our job as experts is to tell clients no, and to get back to the core judgment: who are we building this for, and does this actually make the experience better?”</p><p>Restraint has to come from somewhere, and it won’t come from the tool. “These tools want to cram in as much as possible on every screen,” Rachel said. “This often means we’re drastically simplifying a prototype and stripping out a lot of what was generated. The goal is to surface the essential elements the user <em>actually</em> needs.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/642/0*oAGweaWKJejxcuNP" /></figure><p>Stripping out is expertise, not obstruction. A feature that lives in the wrong place costs you twice: once to build it, and again every time a user has to route around it. The alternative is a bloated system that technically does everything and practically helps no one.</p><p>Rachel flags a related habit slipping across the industry. Teams move so fast now that they skip the step where real users touch the thing. Testing used to be mandatory because building was expensive. Speed made it optional. Optional steps disappear.</p><h3>Bring the Prototype. Bring the Right Expectation.</h3><p>None of this argues against showing up with something you built yourself. The opposite, actually.</p><p>“I’m excited when people come to us with a prototype,” Sarah said. “It saves rounds of back and forth getting to the vision. I can take that starting point, add our professional lens, and level it up.”</p><p>Enterprise teams increasingly work that way. Louisville Water arrives having already sketched what a solution could look like for their business, then hands it over to make real. That starting point is genuinely valuable. It compresses discovery and removes guesswork about intent.</p><p>Chris sees the relationship itself changing shape. “The prototype is becoming more collaborative,” he said. “The customer hands us something they created, and we tweak it, clean it up, and hand it back.”</p><p>Sarah reads the remaining friction as a change management problem rather than a technology problem. The capability arrived faster than the shared vocabulary did, and closing that gap starts with education about what these outputs actually mean.</p><h3>Two Projects, Two Budgets, One Product</h3><p>The confusion makes sense, and it’s fixable. Prototypes look like finished software now, so the old cues that said “early draft” stopped working. A demo shows one path. Production builds the system underneath it and every rule that path never needed.</p><p>The upside is real. That prototype now feeds the build instead of getting thrown away, requirements land clearer, and developers ask a fraction of the questions they used to. Speed went up on both sides of the line. What didn’t change is the line itself.</p><p>So budget for two projects, because that’s what you’re buying. One proves the idea deserves to exist. The other makes it dependable enough that people build their workday around it.</p><p>Anyone can build something that works once. You’re paying for the version that works every time.</p><h3>The Dependency You Don’t Control</h3><p><a href="https://www.yslingshot.com/ai-vendor-risk/"><strong>4 Vendor Moves That Can Break Your AI Product Overnight ⇢</strong></a></p><h3>Written by: Whitney Powell</h3><p>Whitney earned her degree in Marketing and Management from the University of Kentucky and discovered her passion for marketing and events. Her go-getter attitude, willingness to learn, and problem-solving abilities elevate the Slingshot team. Known as a daredevil, Whitney loves trying new things and embracing challenges, whether traveling to new places or taking on new projects at work.</p><p><a href="https://www.linkedin.com/in/whitney-powell-/"><strong>Linkedin</strong></a></p><h3>Expert: Chris Howard</h3><p>Chris has been in the technology space for over 20 years, including being Slingshot’s CIO since 2017. He specializes in lean UX design, technology leadership, and new tech with a focus on AI. He’s currently involved in several AI-focused projects within Slingshot.</p><p><a href="https://www.linkedin.com/in/chris-howard-11084b9/"><strong>Linkedin</strong></a></p><h3>Expert: Sarah Bhatia</h3><p>Sarah Bhatia brings people together. In her decade plus of product and product-adjacent experience, her focus has been on cross-functional collaboration, asking lots of questions, and getting big results. She excels at strategy development, and getting the right brains in the room to solve big problems. Sarah would describe herself as a daredevil, because she’s not afraid to ask foolish questions, get smart answers, and take (calculated) risks.</p><p><a href="https://www.linkedin.com/in/sarahbhatia/"><strong>Linkedin</strong></a></p><h3>Expert: Rachel Foster</h3><p>As UX Lead, Rachel helps guide the design team through strategy, interviews, creation, and testing. Designing software and apps to be both intuitive &amp; beautifully impactful plays into Rachel’s strong desire to connect others. Relying on her Fine Arts background &amp; honed intuition, she gets sudden flashes of ideas and follows them wherever they lead.</p><p><a href="https://www.linkedin.com/in/rachfost/"><strong>Linkedin</strong></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d4939339a932" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[4 AI Learning Tracks That Survive Your Next Tool Switch]]></title>
            <link>https://medium.com/@heyslingshot/4-ai-learning-tracks-that-survive-your-next-tool-switch-4635ccce4fda?source=rss-cc8458c4630f------2</link>
            <guid isPermaLink="false">https://medium.com/p/4635ccce4fda</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[education]]></category>
            <category><![CDATA[learning]]></category>
            <dc:creator><![CDATA[Slingshot]]></dc:creator>
            <pubDate>Sat, 10 Oct 2026 20:41:01 GMT</pubDate>
            <atom:updated>2026-10-10T20:41:01.722Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OZvD6vTuOhibb-prXkGY6w.png" /></figure><p>Your company just paid for AI training. Six months later, half the team switched tools, and most of what they learned went with it.</p><p>That is the quiet math problem inside almost every AI enablement budget right now. Tools shift monthly. Vendors keep offering free half-day sessions on their own AI suites. Leaders keep signing off, checking the box, and assuming the team is covered. Then the stack changes and the training evaporates, because it taught buttons instead of judgment.</p><p>The teams that hold their ground through a tool switch learned something different. They learned how to think about AI, not how to operate one product. And they learned it through their own daily work. Here are the four tracks worth building, and what makes each one survive the next migration</p><h3><em>Summary</em></h3><p><em>The teams that come through a tool switch intact all learned the same thing first, and none of them learned it from a vendor. Judgment is what travels, and it looks different for every role: developers need the boring theory before the agent, designers need enough version control literacy to work inside the pipeline, operations teams need to spot the right problems before they touch a prompt, and leaders need to evaluate rather than approve. Build all four and the next migration costs your team a week of relearning specifics instead of the entire investment.</em></p><h3>Why Vendor Training Expires So Fast</h3><p>Switching tools costs less than most leaders assume. The workflow underneath usually survives the move intact.</p><p>“There’s a lot of overlap in how things get done between tools in the same category,” said Doug Compton, Director of AI Engineering at Slingshot. “You relearn the specifics every time, but you grasp the tool in general quickly, because so much of it carries over.”</p><p>Sarah Bhatia, Director of AI Product Innovation, watched the same pattern when Slingshot’s own team moved off Cursor. “We had to relearn where everything lived,” she said. “But the overall idea of how to build front-end code, how to run that process and iterate on it, we had down pat by then.”</p><p>So the interface is the cheap part. The expensive gap opens when a team never built the underlying process, and the tool was doing the thinking for them.</p><p>This is exactly what a vendor session cannot give you. “Companies bring in their Microsoft partner or their Google partner for a free training, and then they think everyone is trained on AI now,” Sarah said. “Spending half a day hearing from a software rep about which buttons to click might help a few specific use cases. It doesn’t make your team adaptable.”</p><p>Slingshot teaches the opposite way on purpose. “We stay relatively tool agnostic, because the tools switch so quickly anyway,” Sarah said. In practice, that means a room where half the participants use Copilot and half use Claude, with no access to each other’s tools, and a curriculum that has to make both halves effective.</p><h3>Track 1: Developers Learn the Theory Before the Tool</h3><p>The developer track starts somewhere counterintuitive. Not with an agent. Not with a prompt library. With the boring part.</p><p>“They need the foundations first,” Doug said. “Depending on how technical the group is, that means prompting patterns, or software architecture, or how an LLM actually works, so they understand how to interact with it. It might be boring theory, but you build everything else on top of it.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/637/0*9k535MvTGmXgAnB4" /></figure><p>That foundation is what makes the knowledge portable. A developer who understands why context windows degrade output does not need retraining when the window size changes. A developer who only memorized a workflow does.</p><p>Context management earns its own session for that reason. Doug frames it as a half hour with a clear arc: what context is, what breaks when you run out of it, what the side effects look like in real output, and how to run several agents at once without hitting the wall.</p><p>Slingshot also removes the parts nobody should have to rediscover. “We have built-in tool sets so developers get up to speed on running agents effectively,” Doug said. “That stops them from recreating their own setup from scratch.” A weekly internal AI dev session keeps the practices current, and every session ships with a runbook of copy-and-pastable steps. Newer developers follow it line by line. Experienced ones skim to the new parts.</p><h3>Track 2: Designers Keep the Craft, Then Add the Speed</h3><p>Designers stand to gain more reach from AI than almost any role on your team. They also hit the hardest wall getting there, and it has less to do with skill than with habit.</p><p>The technical prerequisites are real but modest. Sarah pointed to branching strategy and pull requests as the baseline that lets a designer work inside an AI-forward pipeline. The bigger differentiator is temperament. “Curiosity and a willingness to learn is a massively impactful skill right now,” she said.</p><p>The resistance comes from craft. “Once you fully embrace AI, it streamlines that UX/UI process and makes all of it much faster,” said Chris Howard, Slingshot’s President. “But it’s hard to give something up when you’ve done it the manual way for years. Designers tend to be perfectionists. Of all the roles, that transition is one of the more challenging.”</p><p>He sees a clear endpoint for the track, though. “The designers who can take an HTML prototype a customer built, bring it in-house, apply standards and good UX practice, and hand it back are not chasing perfection,” he said. “That’s the designer of the future.”</p><p>Skipping that baseline does not get you there faster. It just moves the cost downstream. Sarah already sees the gap in external partners. Designs arrive half generated and half hand-drawn. The discrepancies surface later, when someone has to untangle what the designer actually meant.</p><p>So the design track teaches three things. Enough version control literacy to work inside the pipeline. Practice on inherited work, raising somebody else’s rough prototype to standard. And a stopping rule for what good enough to hand off looks like.</p><h3>Track 3: Operations Teams Start With the Problem, Not the Prompt</h3><p>Your accountant will probably get more out of AI training than your developers will. That sounds backward, but distance from the tools turns out to be an advantage.</p><p>“When I think about our trainings, the people who don’t typically work with AI ended up taking away the most,” Sarah said. “It isn’t something they think about all the time, so they come in ready to reconsider how they work.”</p><p>That happens by design. “We focus first on how to evaluate the kinds of problems where AI is useful,” Sarah said. “We’re pushing people to look at their actual daily tasks and figure out where these tools can be impactful, rather than teaching them how to write a prompt in one specific product.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/632/0*e8wz3Zg1L5G9OvTz" /></figure><p>Chris adds an inventory step that makes the exercise concrete. “Look at the data sources and tools each role already uses,” he said. “Then ask which parts of their day they would most love to get rid of.” From there, you build them something small and safe that removes one of those tasks. “Show them what’s possible with a few examples and they’ll usually pick it up and run with it.”</p><p>Doug’s rule for mixed rooms keeps the track usable at any skill level. Teach to the person who knows the least, but pack in enough that the advanced people still walk out with something.</p><p>So the operations track teaches a sequence, not a toolkit. Spot the problems AI actually fits. Map what each role already works with. Then remove one painful task and let that win pull them toward the next one.</p><h3>Track 4: Leaders Learn to Evaluate, Not Just Approve</h3><p>Evaluating an AI tool used to require a technical reviewer. Now it mostly requires knowing which questions to ask. Leaders who skip that work approve whatever demoed best.</p><p>Chris opens every evaluation the same way. “Start with the privacy policy, so you know where your data is going,” he said. Pull the terms of use, drop them into a chat session that knows your business, and let it surface the clauses that matter. Those policies often change by tier, so the free plan your team piloted may carry different terms than the plan you buy.</p><p>Doug pairs that with a requirements list built before he looks at a single tool. What has to be there, what would be nice, and what the last tool did that this one had better avoid. “Once you have that list, it narrows the possibilities fast,” he said. “And AI can help you build it, and even ask you questions you didn’t know you needed to ask.”</p><p>Every vendor runs a good demo, so Chris looks for what sits behind it. “Any tool that stands behind what it promises will give you a demo period, some real time where you can prove it out.”</p><p>Then comes the hardest call: pulling people out of production work to learn something new. “You need a clear understanding of what you hope to gain as a business,” Chris said. “If you know that, you can usually justify it. But moves like that are costly.” Sarah’s shortcut is to recruit the people already ahead of the curve and let those power users train everyone else.</p><p>So the leadership track teaches evaluation, not approval. Read the terms before the demo. Write the requirements before the pitch. And know what the business gains before you pull anyone off their work.</p><h3>The Real Enablement Strategy</h3><p>Every one of these tracks teaches the same thing in four different accents. Developers get it as theory before tooling. Designers get it as craft before acceleration. Operations teams get it as problems before prompts. Leaders get it as evaluation before approval. The durable part is always the judgment. The tool is just where that judgment gets applied today.</p><p>That is what makes a migration survivable. A team with that judgment loses a week relearning specifics when the tool changes. A team that only learned one product’s menus loses the whole investment and starts over from zero.</p><p>So the question before you approve the next training session isn’t whether your team will learn the tool. It’s whether anything they learn will still be true next quarter.</p><h3>The Terms Changed Since You Signed</h3><p><a href="https://www.yslingshot.com/ai-data-privacy/"><strong>You Are Paying for That Tool. That Does Not Mean Your Data Is Yours. ⇢</strong></a></p><h3>Written by: Whitney Powell</h3><p>Whitney earned her degree in Marketing and Management from the University of Kentucky and discovered her passion for marketing and events. Her go-getter attitude, willingness to learn, and problem-solving abilities elevate the Slingshot team. Known as a daredevil, Whitney loves trying new things and embracing challenges, whether traveling to new places or taking on new projects at work.</p><p><a href="https://www.linkedin.com/in/whitney-powell-/"><strong>Linkedin</strong></a></p><h3>Expert: Chris Howard</h3><p>Chris has been in the technology space for over 20 years, including being Slingshot’s CIO since 2017. He specializes in lean UX design, technology leadership, and new tech with a focus on AI. He’s currently involved in several AI-focused projects within Slingshot.</p><p><a href="https://www.linkedin.com/in/chris-howard-11084b9/"><strong>Linkedin</strong></a></p><h3>Expert: Sarah Bhatia</h3><p>Sarah Bhatia brings people together. In her decade plus of product and product-adjacent experience, her focus has been on cross-functional collaboration, asking lots of questions, and getting big results. She excels at strategy development, and getting the right brains in the room to solve big problems. Sarah would describe herself as a daredevil, because she’s not afraid to ask foolish questions, get smart answers, and take (calculated) risks.</p><p><a href="https://www.linkedin.com/in/sarahbhatia/"><strong>Linkedin</strong></a></p><h3>Expert: Doug Compton</h3><p>Born and raised in Louisville, Doug’s interest in technology started at 11 when he began writing computer games. What began as a hobby turned into his career. With broad interests that range anywhere from snorkeling, science, WWII history and real estate, Doug uses his “down time“ to create new technologies for mobile and web applications.</p><p><a href="https://www.linkedin.com/in/doug-compton-1564438/"><strong>Linkedin</strong></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4635ccce4fda" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Hire, Augment, or Automate: The Decision Most CEOs Make Backward]]></title>
            <link>https://medium.com/@heyslingshot/hire-augment-or-automate-the-decision-most-ceos-make-backward-9005d1f2e017?source=rss-cc8458c4630f------2</link>
            <guid isPermaLink="false">https://medium.com/p/9005d1f2e017</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ceo]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[automation]]></category>
            <dc:creator><![CDATA[Slingshot]]></dc:creator>
            <pubDate>Fri, 09 Oct 2026 20:41:01 GMT</pubDate>
            <atom:updated>2026-10-09T20:41:01.625Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gmRGxjSweQU25zbYAPb51Q.png" /></figure><p>For a long time, a capability gap and a headcount gap were one and the same. Something needed doing, nobody had time to do it, so you wrote a job description. The logic held because no other option cleared the bar.</p><p>Two other options clear it now. You can automate the work, or you can borrow the expertise for a few months instead of hiring it permanently. AI made both cheaper and faster in about two years. Hiring stayed exactly as slow and as permanent as it has always been, and most leaders still reach for it first.</p><p>Telling the three apart before you commit to one is the skill worth building. The signals are clearer than you might expect.</p><h3><em>Summary</em></h3><p><em>Hiring, automation, and augmentation each close a different kind of gap, and the signals that tell them apart are more concrete than most leaders expect. Automation gets cheap in a narrow window, augmentation pays off when the capability stays after the engagement ends, and the durable, specialized work still calls for the right full-time hire, now with AI fluency built into the job. Running the three in the right order is what makes each one land where it actually pays.</em></p><h3>Most Gaps Get Described as a Person, Not a Problem</h3><p>Leaders can answer “who are we hiring” in one sentence. Ask what that person fixes in their first six months and the answer takes noticeably longer to arrive. That second answer decides whether you need a person at all.</p><p>“Historically, it’s just been a role. I need a front-end developer. I need a QA person,” said Chris Howard, President at Slingshot. “We’d get much further if they walked us through the problem they were trying to solve.”</p><p>A job title is a solution wearing the costume of a request. It shows up pre-answered, which is why it clears a leadership meeting so quickly. Nobody argues with a req. The fix here is nearly free. Describe the gap as work rather than as a person, and all three options stay live long enough to compare.</p><p>“When you write out a job description, you’re documenting what the tasks of that job actually are,” said Doug Compton, Director of AI Engineering at Slingshot. “That’s the same starting point you need if you want to make somebody more efficient instead.”</p><p>So the document itself was never the problem. Write it early, read it as an inventory of the work, and it tells you what you could automate and what you could hand to an outside team. Leave it until last, and it just tells HR what to post.</p><h3>The Real Cost of a Hire Starts Long Before Day One</h3><p>Two clocks start when you approve a req, and neither one starts on the new hire’s first day. The first runs from the decision to the signed offer. The second runs from that offer to when the person actually starts the work.</p><p>“The hiring process itself takes months before anyone even starts,” Doug said. “Then the first two weeks, they’re just learning the system, the product, the business logic. It’s usually three to six months before they know the ins and outs of what they’re working on.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/646/0*YcGVgsOmPUMr-QgW" /></figure><p>If you decide in January that you need a senior developer and you won’t start seeing meaningful output until somewhere in the back half of the year. Payroll, meanwhile, started months earlier, and Chris puts benefits and everything else a company provides at roughly 30% on top of salary.</p><p>The harder cost is what those same months buy you if you chose wrong.</p><p>“When you hire somebody, it’s probably the most important move you make as a company,” Chris said. “If you make the wrong move, it takes several months, maybe a year, to figure that out. And then it’s painful to change. Meanwhile you’ve spent a lot of money on someone who didn’t pay off.”</p><p>Both clocks run at the same speed whether the hire works out or not, which is why Chris keeps returning to flexibility as the underrated advantage of outside help. Moving on from a contract stings. Moving on from a bad hire costs far more.</p><h3>AI Raised the Bar on Hiring Instead of Lowering It</h3><p>Everyone gets the same tools. A strong hire uses them to speed up the people around them. A weak one just produces more of what they were already producing.</p><p>“AI has made hiring the right person even more important,” Chris said. “You still want the senior background. But you want them plugged into the AI world, because that combination will accelerate what you already have going.”</p><p>Every role that existed a few years ago now touches AI somewhere, and the people worth hiring know it. “We’re highly unlikely to hire anyone in almost any role who doesn’t at least want to lean on artificial intelligence,” Chris added. “Marketing, sales, software development, design. It’s across the board.”</p><p>He’s also watching entry-level and mid-level roles thin out, including junior developer roles. Those first jobs get harder to land every year. A new graduate who’s gone deep on AI stays interesting in almost any function. One who hasn’t touched it faces a genuinely difficult market.</p><p>Doug flagged a second shift, and it reorders the decision itself. “AI has made everybody more efficient,” he said. “So now we look first at whether we actually need a new person, or whether we can make the people we already have efficient enough to absorb it.”</p><p>Same gap. Same company. Different answer than 2023 would have given.</p><h3>Automation Gets Cheap Where the Work Repeats and the Data Is Ready</h3><p>Two things mark a gap as automatable. The work repeats, and the data behind it is already reachable. Chris starts with repeatability and admits a bias, since he automates whatever he dislikes first. Doug backs that instinct, because clearing the tasks people resent lifts morale and frees experienced people to spend their day on what they actually know.</p><p>Access is the harder test. “Does the AI have access to everything the person doing that job has access to?” Doug asked. “And is it safe to give it that access?” Sometimes the honest answer is no.</p><p>Chris offered an example from inside Slingshot. “We would badly like to turn on the Google Drive connector in Claude,” he said. “It would help a lot of people. But we haven’t done the homework yet to be confident we can do it safely, so it stays off.” The distance between wanting a connector and being ready for one is real work, and it almost never lands on anyone’s roadmap.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/633/0*LwuGKFtVIEv7zHWs" /></figure><p>The vendors won’t flag it for you. “A lot of AI tools default to giving the model access to everything,” Doug said. “They overreach right out of the box, and they give you no good way to rein it in.”</p><p>Which brings up “free,” a word that follows automation around and rarely earns it. A few automations really are close to free, like an inbox monitor that pings someone when a message needs attention. Others hide the bill in custom tools that have to get written and then hosted. Chris adds the line items no vendor will quote you: mapping how the work happens today, retraining everyone around the new workflow, and the real chance somebody leaves over it.</p><h3>Augmentation Buys Expertise Faster Than You Can Hire It</h3><p>Augmentation usually gets framed as a stopgap you run until you can hire properly. The more useful version aims for the opposite outcome: the engagement ends, and the capability stays.</p><p>“If you have a couple of champions inside your organization who just need training or direction, that’s exactly when it makes sense to pull in an outside company,” Doug said. “You’re accelerating their learning, and you’re building that expertise into your own employees.”</p><p>Two conditions make the math work, and AI meets both right now. The skill costs too much to buy permanently, and it moves too fast to grow on your own. “It might be cost-prohibitive to hire somebody who’s an expert in that field,” Doug said. “So you augment, bring in the expertise, and you only have them temporarily.”</p><p>Speed of change is the harder condition to solve by hiring. Whoever you bring on has to track a field that rewrites itself every few months, and they’re doing it between the rest of their job. “An outside company might be living and breathing that technology,” Doug said. “They’re constantly up to date with the latest. You tap into that expertise as needed.”</p><p>Chris framed the payoff differently. “Augmenting with people who know AI can be a catalyst,” he said. “They show your team the way.” A catalyst speeds up a reaction without being consumed by it. That’s the test worth applying before you sign anything, because an engagement that leaves nothing behind was staffing, not augmentation</p><h3>Some Gaps Still Deserve a Full-Time Hire</h3><p>None of this argues for a hiring freeze. Chris hires when the work isn’t going away and doing it well takes real expertise. Both have to be true.</p><p>Doug adds a harder case. “In a very niche area, there isn’t much information for the AI to train on,” he said. “Your expert may simply know more than the model does.” Leading-edge work behaves the same way, because the knowledge is too new for any model to have absorbed it. Hire the person who can reason it out.</p><p>What’s changing is the shape of the role you’re filling. Titles are merging, descriptions cover wider ground, and AI fluency is becoming a requirement across functions. So Chris evaluates candidates against a longer horizon.</p><p>“I think about what this job looks like two years from now,” he said. “You can’t predict the future, but you can feel where things are heading. You want someone who’s going to evolve with it.”</p><p>Doug’s last check happens before you post anything, and it points inward rather than at the market. “Have you done everything you can to train up and support the people you already have?” he asked. “Hiring is a long-term commitment. Do that due diligence first.” Answer that one honestly, and you’ll know whether you’re closing a gap or just adding a person to it.</p><h3>Run the Options in the Right Order</h3><p>Most people never ranked the three options by cost, and that is the whole problem. Hiring takes the longest, commits the most, and corrects the slowest, yet it goes first because a gap instinctively looks like a missing person.</p><p>Flip the sequence. Describe the work instead of the title. Check whether the task repeats and whether your data is ready, since that is the cheapest close when the answer is yes. Borrow expertise when the skill moves faster than you can hire or grow it. Then hire deliberately for the durable, specialized gaps that remain, with a clear picture of what that role becomes two years out.</p><p>The job description sitting in your drafts folder is not the wrong document. It is just answering a question nobody asked yet.</p><h3>Your Team Already Gained Capacity</h3><p><a href="https://www.yslingshot.com/ai-saves-developer-time/"><strong>4 Places AI Buys Back Your Developers’ Time ⇢</strong></a></p><h3>Written by: Whitney Powell</h3><p>Whitney earned her degree in Marketing and Management from the University of Kentucky and discovered her passion for marketing and events. Her go-getter attitude, willingness to learn, and problem-solving abilities elevate the Slingshot team. Known as a daredevil, Whitney loves trying new things and embracing challenges, whether traveling to new places or taking on new projects at work.</p><p><a href="https://www.linkedin.com/in/whitney-powell-/"><strong>Linkedin</strong></a></p><h3>Expert: Chris Howard</h3><p>Chris has been in the technology space for over 20 years, including being Slingshot’s CIO since 2017. He specializes in lean UX design, technology leadership, and new tech with a focus on AI. He’s currently involved in several AI-focused projects within Slingshot.</p><p><a href="https://www.linkedin.com/in/chris-howard-11084b9/"><strong>Linkedin</strong></a></p><h3>Expert: Doug Compton</h3><p>Born and raised in Louisville, Doug’s interest in technology started at 11 when he began writing computer games. What began as a hobby turned into his career. With broad interests that range anywhere from snorkeling, science, WWII history and real estate, Doug uses his “down time“ to create new technologies for mobile and web applications.</p><p><a href="https://www.linkedin.com/in/doug-compton-1564438/"><strong>Linkedin</strong></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9005d1f2e017" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[6 Questions to Answer Before You Turn On Any AI Tool]]></title>
            <link>https://medium.com/@heyslingshot/6-questions-to-answer-before-you-turn-on-any-ai-tool-2d8d823762df?source=rss-cc8458c4630f------2</link>
            <guid isPermaLink="false">https://medium.com/p/2d8d823762df</guid>
            <category><![CDATA[leadership]]></category>
            <category><![CDATA[ai-governance]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[generative-ai-tools]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Slingshot]]></dc:creator>
            <pubDate>Fri, 09 Oct 2026 07:01:17 GMT</pubDate>
            <atom:updated>2026-11-12T16:16:43.214Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8ujUQnULRa7C4ueaKnEAeQ.png" /></figure><p>Turning on an AI tool takes about four seconds: click a toggle, approve a permission screen, and the tool starts working. No ticket or meeting needed. And that speed is the whole appeal, and it’s also the problem.</p><p>Most companies now have something written down about AI, like a policy or a list of approved tools. But the document in your shared directory and the tools running inside your systems this morning may describe two different companies. One is aspirational, and the other has access to your data.</p><p>That gap is where the risk lives. Not in the tools, but in the questions nobody asked before the toggle flipped.</p><h3>Summary</h3><p><em>The answers come down to permissions, interconnects, and containment: split read access from action access before approving anything, review how your approved tools combine rather than evaluating them one by one, and sandbox AI so it only reaches what you explicitly allow. Slingshot’s Steve Anderson and Doug Compton walk through the scoring rubric that turns tool approval from a months-long debate into an hours-long decision, plus the shutoff plan that keeps a failed tool from becoming an outage. The companies getting this right are not the ones with the longest policy document.</em></p><h3>Start With the Data, Not the Demo</h3><p>Almost every AI evaluation starts in the wrong place: ‘what can it do?’ Someone watches a demo and gets excited about the output, features, or possibilities. The better first question is far less interesting but far more useful.</p><p>“One of the first things I look for is what data it’s accessing,” said Steve Anderson, Principal Developer and AWS Architect at Slingshot.</p><p>Doug Compton, Slingshot’s Director of AI Engineering, framed it as two questions. “The data is the big one, and then what can it do with that data?”</p><p>Those are separate risks. Reading your customer records is one thing. Sending an email on your behalf is another. Most tools let you split those permissions apart, and most companies never look closely enough to do it.</p><p>“With Gmail, you can tell AI to be read-only, or you can allow it to draft an email for you,” Doug said. “You can give the AI actions it’s allowed to take. But a lot of the time, those permissions just don’t go far enough.”</p><p>The scope also tends to be wider than people expect. “At Slingshot, when we looked at hooking up Google Drive to the OAuth application, the permission set was shockingly broad,” Steve said.</p><p>So when your team brings you a tool, ask to see the permission screen rather than the vendor’s feature page. Those two documents rarely tell the same story.</p><h3>Safe Tools Can Add Up to an Unsafe System</h3><p>Here is the part that surprises most leaders: you can approve every tool individually, correctly, and still end up exposed.</p><p>“One connector you enable might give AI access to sensitive data, and then another tool you enable gives it the ability to do something with that data, possibly making it public,” Doug said. “So you have to look at the interconnect between these tools as well.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/631/0*tHw83HP6al3_GLJo" /></figure><p>Tool A reads. Tool B publishes. Neither one is dangerous alone, but together they’re a leak with a schedule, and reviewing tools one at a time misses it entirely.</p><p>The same thing happens over time. A tool you approved in March isn’t the same tool in October, and new capabilities arrive enabled by default. “We’ve seen AI tools push updates that were opted in by default,” Steve said.</p><p>His fix is unglamorous yet effective. “Whenever there’s a release, you have to check what the release notes say before you install that update. And you should have a policy that dictates the cadence you’re reviewing your vendors.”</p><p>Annual vendor review made sense when software changed annually. AI tooling doesn’t run on that calendar.</p><h3>“Nobody” Is the Most Expensive Answer</h3><p>Ask who owns your AI tools. If the honest answer is that no single person does, you’ve found your most urgent gap.</p><p>“That’s the gap a lot of IT teams are dealing with right now: being able to manage AI tools and keep them locked to policies,” Steve said. “Your IT team should be the one ensuring whatever tools are in place are approved and operating within your policies.”</p><p>Central control helps, but it has a ceiling. Doug walked through where it stops: “Admins can enable and disable tools from a centralized location. But that’s really just restricting the desktop app. Somebody can still install a tool locally, or reach one across the internet. Disabling a tool in the admin console doesn’t mean the AI can’t still get to sensitive information.”</p><p>That is shadow AI, and it’s not hypothetical. “Shadow AI is a very difficult thing that people are struggling with right now,” Steve said. You cannot monitor your way out of it, either. “Without putting spyware on everything, you have to trust your people to an extent to follow the policies you put in place.”</p><p>So your policy only works if your people understand it well enough to follow it, which is a training problem long before it’s a technology problem.</p><h3>Decide What Sits Between the Tool and the Outside World</h3><p>One question separates a contained mistake from a public one. When this tool produces something, who sees it before your customer does?</p><p>Both Doug and Steve have watched what happens when the answer is nobody. “The worst thing I’ve had AI do is automatically update AWS in our dev environment,” Doug said. “I was watching it, so it wasn’t a big deal. But it raised my eyebrows at what it’s capable of.”</p><p>Steve’s example was quieter and more instructive, because nothing broke. “I enabled AI to create Jira tickets based on reviews and end-to-end testing automatically. Nothing awful, but the volume of tickets and the nitpicky level became a flood of information. It made the project difficult to navigate.”</p><p>The system did exactly what it was told. It just buried a team in noise, which is its own kind of failure.</p><p>Better instructions help, but they don’t guarantee much. “A good prompt around the tools alleviates a lot of that, but it’s trial and error,” Doug said. “You don’t know you need that section of the input until the tool does something stupid or dangerous. And the prompt still doesn’t stop it from doing it again; it just makes it harder.”</p><p>And your non-technical teams can’t catch what they were never taught to look for. “This is so new and so different from what we’ve had to deal with in the past that companies really need to give their employees training just to know what’s possible and what to look out for,” Doug said. “Some of this isn’t obvious.”</p><h3>Contain It Before You Have to Kill It</h3><p>Most companies never consider a containment layer, and it does more work than another paragraph of policy ever will.</p><p>“You need to contain AI in a sandbox, so it only has access to the things you allow,” Doug said. “It doesn’t need access to your whole computer, your documents, or the network unless you explicitly say so. Sandboxes are almost like virtual machines. You control exactly what files it has access to, and you can put a firewall between it and the internet.”</p><p>Containment also buys you an exit, since it’s more than likely that a tool will one day change, break, or fail a review. “Hopefully you had a shutoff plan to begin with,” Steve said. “If you’ve built your entire application around the AI, it’s a single point of failure. If you have to shut things down while you fix it, you’re just going to be down.”</p><p>Decide the borders while the tool is still optional, and the day you have to pull it stays a decision, not an outage.</p><h3>A List of Approved Tools Is Not Governance</h3><p>Most AI policies stop at a list: use only the tools leadership signed off on. The list is a fine start and a poor finish, for two opposite reasons.</p><p>It goes stale faster than anything else in your stack. “That list of tools is moving and changing quickly, much more than other types of software,” Doug said. “If you want to get the most out of AI, you need to look at that approved list more frequently. You’d hate to miss out on something that would help the company in a significant way.”</p><p>A list also tells your team what was decided without telling them how to decide. Steve’s answer is a guide. “You want a scoring rubric for whether or not you should approve a tool. Think of things you should look at when evaluating: what are the things that are okay for a tool to do, and what are our hard lines where we won’t approve it without particular remediations in place?” Write the rubric down once, and every future tool gets evaluated in hours instead of months.</p><p>Skipping that work costs more than it looks. “There are a million stories of people who deleted production databases because they weren’t careful,” Steve said. Doug pointed at the quieter version. “Your company could send out a bunch of emails written by AI that were wrong or embarrassing, and now the world knows you did stupid stuff.”</p><p>Those are mistakes you make. Your intellectual property is exposure you may never see. When <a href="https://www.theguardian.com/science/2026/sep/08/openai-claims-to-have-solved-maths-problem-that-stumped-humans-for-decades">OpenAI announced in September 2026 </a>that its agents had cracked part of the Navier-Stokes problem, mathematicians working on the adjacent problem publicly questioned whether their research data, run through OpenAI’s Codex tool, had fed the effort. OpenAI denied accessing their work and said no specific user data was involved, while recognizing it couldn’t fully rule out an indirect connection.</p><p>Scale is what makes that dangerous for a company your size. “If you don’t have a big enough data population, you can’t de-identify things,” Steve said. If your company is the only one in the world doing something a particular way, stripping your name off it protects almost nothing. He landed the whole question in one comparison. “Just like you wouldn’t give your intern the keys to the kingdom, you shouldn’t give them to AI either.”</p><h3>The Toggle Is the Easy Part</h3><p>None of this argues for slowing down; it argues for knowing what you just agreed to.</p><p>The companies getting AI right are not the ones with the longest policy document. They are the ones who can answer a short list of questions about every tool they run: what data it touches, what it can do with that data, what other tools it can reach, who owns it, who checks its work, and how you turn it off on a bad Tuesday. That is governance. The policy is just where you write it down.</p><p>Your team can turn on an AI tool in four seconds. The only real question is whether anyone can tell you what happened in the next four.</p><h3>Where Policy Meets Training</h3><p><a href="https://www.yslingshot.com/ai-learning-tracks/">AI Learning Tracks To Survive A Tool Switch ⇢</a></p><h3>Written by: Savannah Cherry</h3><p>Savannah leads marketing and new business at Slingshot. She writes, posts, and creates all things Slingshot, and helps companies navigate working with a tech partner for the first time. While she isn’t developing software, her CIS minor and a tendency to tinker with AI tools to streamline her own work keep her up to speed on the team’s work. She co-organizes and hosts the Louisville AI Exchange, and she can’t rest until all her work is done.</p><p><a href="https://www.yslingshot.com/author/savannah/">View All Posts</a></p><h3>Expert: Doug Compton</h3><p>Born and raised in Louisville, Doug’s interest in technology started at 11 when he began writing computer games. What began as a hobby turned into his career. With broad interests that range anywhere from snorkeling, science, WWII history and real estate, Doug uses his “down time“ to create new technologies for mobile and web applications.</p><p><a href="https://www.linkedin.com/in/doug-compton-1564438/">Linkedin</a></p><h3>Expert: Steve Anderson</h3><p>Steve is one of our AWS certified solutions architects. Whether it’s coding, testing, deployment, support, infrastructure, or server set-up, he’s always thinking about the cloud as he builds. Steve is extremely adaptable, and can pick up the project and run with it. He’s flexible and able to fill in where needed. In his spare time, he enjoys family time, the outdoors and reading.</p><p><a href="https://www.linkedin.com/in/steve-anderson-5b13bb1/">Linkedin</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2d8d823762df" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>