<?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 Michael Robinson on Medium]]></title>
        <description><![CDATA[Stories by Michael Robinson on Medium]]></description>
        <link>https://medium.com/@michael_robinson?source=rss-1d37ba781dbe------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*PdqoU1Ff3UbeESOc_ZzKJQ.png</url>
            <title>Stories by Michael Robinson on Medium</title>
            <link>https://medium.com/@michael_robinson?source=rss-1d37ba781dbe------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 18:21:42 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@michael_robinson/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[Our Partnership with ASI: America’s engine for critical industries]]></title>
            <link>https://medium.com/@michael_robinson/our-partnership-with-asi-americas-engine-for-critical-industries-0656dbcfe175?source=rss-1d37ba781dbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/0656dbcfe175</guid>
            <dc:creator><![CDATA[Michael Robinson]]></dc:creator>
            <pubDate>Wed, 21 Oct 2026 17:04:07 GMT</pubDate>
            <atom:updated>2026-10-21T17:05:42.764Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Investing in ASI’s $325 million Series C to build the decision making engine for the world’s most critical industries.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qCaNsiKpI8o4k4cxg9BOsw.jpeg" /></figure><p>The world’s most critical infrastructure has become too complex for the software it runs on. Across aviation, defense, energy, and logistics, operators are responsible for thousands of interconnected assets in increasingly dynamic environments, yet still rely on outdated software and fragmented systems to manage them. They’re left piecing together information manually and reacting to problems after they’ve already started, with limited ability to anticipate what comes next or how best to respond.</p><p>Advances in AI create the opportunity to fundamentally change that equation. Phillip Buckendorf and his team at ASI are building toward that future. Their ambition is to build the operating intelligence behind the world’s most important physical systems, what Phillip calls the “automated nervous system” for critical infrastructure.</p><p>When we first met Phillip in 2025, we were immediately struck by the scale of his ambition and his decision to tackle one of the most complex and high-stakes industries first: aviation. Safety is paramount, reliability is nonnegotiable, and even small disruptions can cascade across thousands of flights and passengers. Phillip began by deploying ASI’s technology with major airlines, including Delta, Alaska, and United, proving its capabilities in real-world operations. That commercial foundation helped ASI win the FAA’s $875 million program of record to modernize air traffic management, beating Palantir and Thales. ASI is one of only four technology companies founded since 2000 to win a federal program of record of this magnitude, joining Palantir, SpaceX, and Anduril. The scale of the challenge is enormous. Every day, the National Airspace System manages roughly 45,000 flights and 2.9 million passengers. The image above shows ASI’s SMART system in Washington, which analyzes flight schedules, weather, and airspace capacity to anticipate congestion and recommend adjustments before delays cascade across the network.</p><p>As we spent more time with Phillip, it became clear that aviation was just the beginning. The same technology that helps airlines anticipate disruptions and optimize aircraft movements can be applied to virtually any complex physical system. ASI’s platform continuously models how thousands of interconnected assets interact, predicts how changes ripple through a network, and helps operators make better decisions in real time. Whether coordinating aircraft, military operations, or electric grids, the underlying challenge is remarkably similar: managing enormous complexity, anticipating what happens next, and optimizing outcomes.</p><p>We spoke with customers across ASI’s core industries who consistently praised the exceptional team, the strength of the technology, and its game-changing impact on their operations. That customer enthusiasm is matched by significant commercial and government traction. ASI is already working with the U.S. Army, U.S. Air Force, and U.S. Pacific Command. The Army recently awarded ASI a $99 million contract to scale its technology across modernized command and control efforts, including the Next Generation Command and Control program. In energy, National Grid has partnered with ASI on electric grid planning and operations.</p><p>The combination of Phillip’s leadership, the exceptional customer feedback, and demonstrated execution gave us deep conviction that ASI is the company to build the decision making engine for the world’s most critical industries. We’re proud to have invested in ASI’s $325 million Series C alongside Spark Capital, Meritech Capital, Andreessen Horowitz, Energy Impact Partners, and Citadel Securities. We’re thrilled to back Phillip and the entire ASI team as they fundamentally transform how the world’s critical infrastructure operates.</p><p>Written by: <a href="https://medium.com/u/1d37ba781dbe">Michael Robinson</a>, <a href="https://medium.com/u/4f281508f4de">Kindle Van Linge</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0656dbcfe175" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Our Partnership with Noetive: Building the Intelligence of Record for the Physical Economy]]></title>
            <link>https://medium.com/@michael_robinson/our-partnership-with-noetive-building-the-intelligence-of-record-for-the-physical-economy-b7aa5c14d773?source=rss-1d37ba781dbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/b7aa5c14d773</guid>
            <dc:creator><![CDATA[Michael Robinson]]></dc:creator>
            <pubDate>Tue, 06 Oct 2026 00:40:28 GMT</pubDate>
            <atom:updated>2026-10-06T00:46:48.052Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Bringing Self-Improving AI to the $30 Trillion Physical Economy</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wx9Sy3G1dhvCDXbnO_iUZA.jpeg" /></figure><p>AI learned the digital world from the internet. Code, text, images, and decades of digital activity gave models the raw material to transform knowledge work. The physical economy never had that advantage. Factories, warehouses, fleets, plants, and data centers run on information that largely exists nowhere else. The record lives in three places software never unified: what is actually happening on the floor, the company’s own systems, and the operators who already know how the pieces fit. That is a $30 trillion slice of global GDP that AI has barely touched. The company that learns how this world actually operates, and gets smarter with every operation, becomes the intelligence layer the physical economy runs on. We are partnering with Noetive to build it.</p><p>Craft invested in the <a href="https://noetive.ai/news/noetive-emerges-from-stealth-with-41-million-seed">$41 million seed round</a> led by Eclipse alongside The Westly Group, Swish Ventures, Factory, Incite Ventures, Gigascale Capital, Operator Partners, Liquid 2 Ventures, and individual investors including Meta CTO Andrew Bosworth, Airbnb CTO Ahmad Al-Dahle, Nest co-founder Matt Rogers, and Zach Frenkel.</p><p>A big part of our conviction comes down to Amir, Dan, and the founding team. Amir and Dan have known each other for more than twenty years. Amir spent a decade at Meta across GenAI and XR, after stints at Google, Amazon, and HP. Dan founded and sold Stackbit to Netlify after product roles at Lyft and Meta. That combination matters here. Noetive has to push the frontier of what AI can understand about the physical world while turning that intelligence into products operators trust to run their businesses. Amir, Dan, and the team bring the technical depth, product instinct, and history together to do both.</p><p>Physical industries don’t lack software. They’ve been buying ERP, MES, WMS, and other systems for thirty years. The problem is that the data is fragmented, and much of the most important information has never been captured. Existing systems each see a piece of the operation, but none can reason across the whole thing: labor, materials, capacity, schedules, downstream commitments, and what is actually happening on the floor. So the job of putting it all together still falls to people. Every industrial business relies on a handful of operators who have internalized how the whole operation fits together. Their judgment is one of the company’s highest leverage assets and one of its biggest risks. It doesn’t scale, walks out the door every night, and eventually leaves for good. We’ve seen this pattern before: expert judgment that creates enormous value but can’t be systematized, scaled, or retained. We believe Noetive can change that for the physical economy.</p><p>Noetive’s self improving AI learns how an industrial business actually runs, makes decisions, observes what happens next, and learns from the outcome. But to understand an operation, it needs to see more than what exists in software. Noetive pairs its AI with sensing hardware inside the operation, so the system learns from the work itself, not only from the software that describes it. This closes the loop between the plan and what actually happened, turning every operation into a new learning signal. Noetive calls the result “the intelligence of record.” The system of record was the last generation’s defensible position in enterprise software: whoever held the data held the account. The intelligence of record is the future. The system that makes the decisions, learns from every outcome, and gets better every week becomes something the business can’t run without.</p><p>The early deployments are already producing better decisions in real time, and at a level of precision that was previously impossible. Noetive continuously optimizes across labor, materials, capacity, and changing conditions. The result is a fundamentally more efficient operation, with improvements that show up directly in revenue and margin. The system does not stop there. Every operation is a training signal, every outcome sharpens the model, and every new customer makes the next one faster to learn. That compounding loop is the bet: that the company that learns how the physical economy actually runs, and keeps learning every time it runs, becomes the intelligence layer a $30 trillion economy can’t run without.</p><p>We’re very bullish on the convergence of AI and the physical world. We believe Noetive has the power to capture this, and we’re grateful to be along for the journey.</p><p>Written by: <a href="https://medium.com/u/1d37ba781dbe">Michael Robinson</a> and <a href="https://medium.com/u/4e969cb516d1">Taylor Durand</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b7aa5c14d773" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Triple Down on Horizon3: The Autonomous Security Platform for the “AI vs AI” Era]]></title>
            <link>https://medium.com/@michael_robinson/triple-down-with-horizon3-the-autonomous-security-platform-for-the-ai-vs-ai-era-633e8ca460a5?source=rss-1d37ba781dbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/633e8ca460a5</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Michael Robinson]]></dc:creator>
            <pubDate>Tue, 11 Aug 2026 12:25:37 GMT</pubDate>
            <atom:updated>2026-08-11T13:48:20.132Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rGySWDa3dCq9-l8uZ2ki7Q.jpeg" /></figure><p>We’re excited to share that Craft Ventures has invested in <a href="https://horizon3.ai/news/press-release/horizon3-raises-250m-series-e-at-2b-valuation-to-lead-the-ai-vs-ai-cybersecurity-era/">Horizon3’s $250 million Series E</a> at a valuation of more than $2 billion, co-led by NightDragon and NEA. Craft first partnered with Horizon3 in July 2023 when we led the company’s Series C, and this marks our third significant investment in the company.</p><p>A lot has changed in the world of AI and cybersecurity in only three short years, but Horizon3 accurately predicted where the world was heading: machine-speed hacking, leading to algorithms fighting algorithms, driving the need for autonomous security. Snehal brought a uniquely informed perspective to that future, having led technology on both the defensive and offensive sides of cybersecurity as CTO of Splunk and later CTO of Joint Special Operations Command. He had seen firsthand how quickly attack and defense capabilities were evolving, and more than 20 CISOs we spoke with reinforced the same conclusion: as attacks became more automated, defense had to become continuous and autonomous. We shared that conviction and partnered with Horizon3 to build the world’s best AI hacker to power a proactive security platform.</p><p>When we first invested, frontier models didn’t have notable cyber capabilities and AI hacking was not yet in the mainstream. At the time, we wrote that continuous security management would be the next evolution in cybersecurity and believed Horizon3’s proprietary cyber terrain maps would compound with every test, powering a broader security platform over time. Three years later, both parts of that thesis are playing out.</p><p>Today, AI hacking is the white-hot center of both frontier AI and the cybersecurity industry. Post-Mythos models can quickly discover thousands of vulnerabilities and chain low and medium findings into workable exploits. The gap between discovery and real-world exploitability is effectively zero. These capabilities are also more accessible than ever before as the latest open-source models are catching up to frontier labs. In this new world, understanding what is actually exploitable across your organization has become an even more urgent priority.</p><p>The rapid advance of AI has created a massive inflection in the market opportunity for Horizon3. Many thought AI hacking was just a tool for red-teamers, a useful but relatively narrow market. The idea that AI hacking would reshape exposure management was still a hypothesis three years ago, but security teams have realized that the only credible way to understand their exposure is to continuously test their defenses the way a real world AI-enabled attacker would. The same shift is happening in application security, where budgets are moving from DAST to agentic penetration testing. The market need has shifted from vulnerability discovery to continuous validation: prove what’s exploitable, fix it, verify the fix, and repeat.</p><p>For those reasons, the market of AI hacking has attracted many new startups and venture capital dollars. At BlackHat, there were more than 20 companies mentioning autonomous penetration testing. Over the last two years, there has been over $1 billion in funding across these companies. Horizon3 makes up $350 million, or 35%, of the total capital raised. Horizon3 has captured the lion’s share for a number for reasons, chiefly due to their solution breadth, purpose-built AI architecture, real-life training data accumulated over six years of testing, and the trust that comes along with that experience:</p><p><strong>Solution breadth</strong>: There is confusion in the market about AI hackers. Most only cover applications. Horizon3 covers the full attack surface across infrastructure, identities, and applications. They are also on the frontier of LLM-based attack defense, specifically around deception techniques. The next paradigm is extending into autonomous defense, closing attack paths with verification running as one continuous learning loop between AI attackers and AI defenders.</p><p><strong>Purpose-built AI architecture: </strong>Horizon3 combines frontier models with other AI techniques to get large performance gains at a fraction of the cost of token-hungry alternatives, and cost is what determines whether continuous validation is actually viable. Many competitors cannot run continuously or they require human augmentation.</p><p><strong>Real-life cyber terrain dataset: </strong>Every one of Horizon3’s 310,000 production pentests adds to their Cyber Terrain map, a dataset of real-world cyber terrain that Horizon3 has been accumulating for six years. That loop between attack and defense can’t be recreated with generic models or public data.</p><p><strong>Trusted by leading organizations: </strong>Horizon3 is trusted to protect over 7,000 organizations globally, including four Fortune 10 enterprises, multinational banks, major healthcare networks, and large classified government agencies. They are actually trusted to run in production because they’ve achieved zero downtime across all production tests, with no humans in the loop.</p><p>Horizon3’s North Star has always been to build a closed-loop system where the world’s best AI hacker finds problems, fixes them, and continuously verifies security. Frontier AI has made the attacker’s job easier than ever, and the only adequate response is proactive defense that operates at the same speed and scale. Horizon3 has been building exactly that, and we’re excited to continue backing the team as they define the future of autonomous security.</p><p><em>Written by: </em><a href="https://medium.com/u/1d37ba781dbe"><em>Michael Robinson</em></a><em>, </em><a href="https://medium.com/u/e3cb4ee22379"><em>Kevin Gabura</em></a><em>, </em><a href="https://medium.com/u/da8fff35d560"><em>Dan Moor</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=633e8ca460a5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Our Partnership with Neo: Secure Agentic Software]]></title>
            <link>https://medium.com/@michael_robinson/partnering-with-neo-to-secure-agentic-software-43d2dae5740c?source=rss-1d37ba781dbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/43d2dae5740c</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[venture-capital]]></category>
            <dc:creator><![CDATA[Michael Robinson]]></dc:creator>
            <pubDate>Tue, 21 Jul 2026 17:02:10 GMT</pubDate>
            <atom:updated>2026-10-06T00:09:18.832Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ksdw-DfeTq30euBAZhvf2w.jpeg" /></figure><p>As Neo emerges from stealth, we’re excited to share Craft’s investment in the <a href="https://www.neo.ai/news/neo-launches-100m">$100M round</a> alongside Andreessen Horowitz, Bessemer Venture Partners, and Merlin Ventures. Neo is building the <strong>Agentic Software Control </strong>platform, a real-time control layer to secure organizations’ software environments that are becoming agentic. We are proud to have backed Neo from the earliest stage and look forward to deepening our partnership this round with a significant investment.</p><p>Our conviction in Neo began with Nick Warner. Three years ago, Nick became a cybersecurity advisor to Craft. Since then, he has been an extraordinary partner to our firm and portfolio. Nick deeply understands how the most consequential cyber categories are won and lost. That knowledge was earned after a decade in the EDR market, one of most unforgiving and competitive battlegrounds in cybersecurity. He held senior roles at McAfee and Cylance and went on to build the go-to-market machine that took SentinelOne from $1 million to $500 million of ARR and led them through IPO as COO and President.</p><p>When Nick was ready to start a company, we were eager to partner with him and his co-founders, Shlomi Salem and Eran Shirazi. Shlomi led detection engineering at SentinelOne for over a decade and Eran built EasySend, the customer experience company. The three of them held a shared view around how software capabilities were expanding with AI and how quickly that was going to overwhelm security teams. According to Gartner, only 5% of enterprise applications featured agentic capabilities in 2025, but 40% will by the end of 2026. Everyday apps that enterprises already use and trust are gaining the ability to reason, reach into other systems, call tools, and carry out actions with legitimate credentials.</p><p>Security teams are spending tens of billions of dollars per year securing their software, but none of it covers this new layer of agentic capabilities. After spending months speaking with CISOs in our community about this challenge, it became obvious that this is a critical gap they need to fill in order to enable broad AI adoption in their organizations. Neo addresses this by providing a real-time control plane that covers all the software running across the enterprise. Neo sees what agentic software can do, understands how it behaves, and enforces policy inside the software layer before risky activity happens.</p><p>The velocity at which Neo builds is remarkable. They began by developing a scanner that generates an organization’s full agentic software inventory in minutes. They then enrich those inventories with their proprietary “Neoverse,” a knowledge base of the real capabilities and risks of the entire agentic software landscape, covering over 1.2 million skills, binaries, MCP packages, and third-party AI models. The Neo platform is powered by their advanced endpoint sensor, which captures every prompt and action on the device and understands the intent behind it. That architecture enables them to enforce guardrails natively inside the software layer without handoffs to other tools. They also developed an LLM-powered policy creation engine that creates guardrails autonomously for SecOps teams. In a crowded and often confusing AI security market, the Neo team is very clear and explicit on solving the agentic software control problem and building a deep platform there.</p><p>The market potential for agentic software control is similar in size and scope to the cloud security market, which was at its infancy a decade ago and is now a $20 billion opportunity. The key difference is that the adoption curve for AI-enabled software will be much steeper than it was for cloud. Traditional software is rolling out these capabilities fast, without express permission from security teams, and employees are adopting AI-native tools en-masse.</p><p>Some of the largest enterprises have already deployed Neo to secure their agentic software layer. They realize the agentic train has left the station and managing the inevitable risk associated with that will require endpoint-based visibility and control that covers all agentic software capabilities. Every week we are seeing developments in agentic capabilities, which is simply too fast for the largest cyber vendors to keep up with their platforms.</p><p>We are excited to partner with Neo as they become the go-to agentic software control company. They have an experienced team that has built and scaled enterprise-grade sensor-based platforms before, and they are quickly becoming critical infrastructure for agentic software adoption.</p><p><strong>A new layer of agentic software deserves a new layer of control. That’s Neo.</strong></p><p><em>By </em><a href="https://medium.com/u/1d37ba781dbe"><em>Michael Robinson</em></a>, <a href="https://medium.com/u/e3cb4ee22379">Kevin Gabura</a>, <a href="https://medium.com/u/da8fff35d560">Dan Moor</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=43d2dae5740c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Partnering with Ent to Protect Work as It Happens]]></title>
            <link>https://medium.com/@michael_robinson/partnering-with-ent-to-protect-work-as-it-happens-3d19facd2a6b?source=rss-1d37ba781dbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/3d19facd2a6b</guid>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <dc:creator><![CDATA[Michael Robinson]]></dc:creator>
            <pubDate>Tue, 21 Jul 2026 17:00:52 GMT</pubDate>
            <atom:updated>2026-07-21T17:00:52.509Z</atom:updated>
            <content:encoded><![CDATA[<p><em>By Michael Robinson, Kevin Gabura, and Dan Moor</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Aeey968rsYOBBoLj-GLrRw.jpeg" /></figure><p>We’re excited to announce Craft’s investment in Ent’s $100M round led by Decibel, with participation from Sequoia, Crosspoint, Shield, Felicis, and IQT. This partnership is especially meaningful to us because our relationship with Lou Manousos goes back a decade. Lou is the kind of founder you back in a heartbeat: an exceptional technologist, product thinker, and operator who has already built and scaled a category-defining cybersecurity company.</p><p>We first got to know him when he and Brandon Dixon were building RiskIQ, the attack-surface management company. After RiskIQ was acquired by Microsoft, Lou and Brandon helped lead the development of Security Copilot. That vantage point gave them a unique view into how hardware, foundation models, and operating systems were converging. They saw the conditions emerging to run language models directly on the endpoint. That insight became the genesis of Ent.</p><p>When Lou and Brandon shared their vision for Ent, the opportunity was clear. On-device language models could unlock security applications that were not possible before. The last generation of endpoint security companies created hundreds of billions of dollars of value. Ent is bringing that same kind of opportunity to the AI era, with new capabilities powered by the latest defensive AI models. We jumped at the opportunity to partner with them and believe the next generational cybersecurity company will be built where work actually happens.</p><p>This technology is arriving at a moment when organizations urgently need it. Frontier AI is changing how people work, expanding the enterprise attack surface, and accelerating how quickly attackers can move. Enterprises are embracing a world where humans and AI work together to improve productivity. Security teams do not want to slow that progress down. At the same time, threats that once unfolded over days can now happen in seconds. In this world of AI-driven workforces and AI-powered adversaries, reactive security tools are no longer enough. Enterprises need the ability to prevent risk in real time, not react after an alert.</p><p>Ent helps organizations do this with an intent-aware workspace security platform that operates directly on the device. The platform acts like an always-on security companion. It protects users and AI agents as work is happening. By bringing specialized defensive AI reasoning to the endpoint, Ent helps customers move from reactive detection to real-time prevention. Because inference runs locally, Ent can act with effectively zero latency while keeping sensitive data on the endpoint.</p><p>Building Ent required solving a serious technical challenge. The team needed to build a lightweight agent capable of running AI models without taxing a device’s CPU or memory. They needed the AI to build a “work model” by observing how a company’s human and AI workers operate. That model helps Ent understand the intent behind actions and distinguish normal work from risky behavior. They also needed a great user experience because Ent interacts with workers in real time through natural language nudges and action blocks.</p><p>To make this possible, Lou and Brandon assembled a world-class team made up almost entirely of principal-level engineers and threat researchers from leading cybersecurity companies and U.S. intelligence agencies. That depth of technical talent shows up in the product. Ent’s lightweight agent uses less than 1% CPU and under 100MB of memory, while supporting use cases like insider risk detection, data loss prevention, and AI governance. The platform is already trusted by Global 2000 customers across hospitality, financial services, and defense, with more capabilities launching soon.</p><p>We could not be more excited to deepen our partnership with Lou, Brandon, and the Ent team as they come out of stealth. Ent is bringing the latest defensive AI models to the endpoint and helping organizations move toward real-time prevention. We look forward to supporting the team on their journey to protect work as it happens.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3d19facd2a6b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[All in on Supabase: the Default Backend for Software Development]]></title>
            <link>https://medium.com/@michael_robinson/all-in-on-supabase-the-default-backend-for-software-development-609169c4a523?source=rss-1d37ba781dbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/609169c4a523</guid>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[coding]]></category>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[software-development]]></category>
            <dc:creator><![CDATA[Michael Robinson]]></dc:creator>
            <pubDate>Thu, 04 Jun 2026 17:40:25 GMT</pubDate>
            <atom:updated>2026-06-04T18:12:10.397Z</atom:updated>
            <content:encoded><![CDATA[<p><em>AI is causing an explosion in software creation and Supabase has become the default backend for this new era.</em></p><p>By <a href="https://medium.com/u/1d37ba781dbe">Michael Robinson</a> &amp; <a href="https://medium.com/u/d7706d1a9fd9">Zao Chen</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mxNFS5Q_DPxEzhFxX7-RsQ.jpeg" /><figcaption>Craft Ventures invests in Supabase’s Series F, making it our fourth investment in the company.</figcaption></figure><p>Coding has entered a new era of abundance. Millions of people from developers to non-technical users are harnessing the power of AI with tools like Claude Code and Cursor to build applications at an unprecedented pace. But applications don’t run themselves. Behind every application lies infrastructure to manage data, users, and workflows so everything runs smoothly as it scales. The AI software boom needs a backend built for speed, scale, and developer love and that backend is Supabase. Supabase removes the heavy lifting to spin up a backend, making it easier for builders to ship production-ready applications faster. The excellent developer experience that made Supabase wildly popular since its inception has naturally led it to become the backend of choice for AI-powered coding tools and AI native development workflows.</p><p>Today, nearly 10 million developers use Supabase, almost a 4x increase from just a year ago. That number continues to grow as builders join the community and contribute to the Supabase platform. Even as Supabase has scaled to millions of developers, the company has preserved what made it special from the start: an obsessive focus on developers. To this day, co-founders Paul and Ant remain actively engaged with the community, regularly answering questions and responding to comments on X and Hacker News (no, it’s not an AI bot).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XSluzNcr5hzxglB0QAWnXg.png" /><figcaption>Supabase’s user growth since inception</figcaption></figure><p>AI has unleashed an explosion in application development. The next frontier is scale. As AI-built applications move from prototypes to production, Supabase will need to support some of the largest database workloads in the world. That is why we are so excited about Multigres. Multigres is a horizontally scalable Postgres architecture designed for enterprise-grade and massive scale database workloads. To lead its development, last year Supabase brought on Sugu Sougoumarane, one of the creators of Vitess at YouTube. Just as Vitess helped MySQL power some of the world’s largest applications, Multigres has the potential to do the same for Postgres. It will enable developers building on Supabase to scale from millions to billions of users, and eventually to the trillions of AI agents that will define the next era of software.</p><p>Since co-leading Supabase’s Series C in 2024, Craft has invested in every round — Series D, E and now series F. Even though the phrase is “doubling down,” we’re actually “quadrupling down.” With each round, our conviction has only grown and we’re deeply grateful to be working with this incredible team. This Series F financing represents Craft’s largest investment in Supabase to date and we’re just getting going. Supabase has firmly established itself as one of the most important companies in AI-native software, setting the standard for how the next generation of applications are built, shipped, and scaled. Congratulations to the incredible team and people at Supabase for this incredible milestone. We couldn’t be more excited and proud to be on the journey with such a legendary team and company.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=609169c4a523" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Our Partnership with Oasis Security: Securing Access for the Agentic Era]]></title>
            <link>https://medium.com/@michael_robinson/our-partnership-with-oasis-security-securing-access-for-the-agentic-era-106e0b1cfbee?source=rss-1d37ba781dbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/106e0b1cfbee</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Michael Robinson]]></dc:creator>
            <pubDate>Thu, 19 Mar 2026 20:01:16 GMT</pubDate>
            <atom:updated>2026-03-19T21:05:16.661Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bWu2LimQDuyoGKgGgqcqbw.jpeg" /></figure><p>We are excited to share that Craft Ventures is leading the $120 million Series B in <a href="https://www.oasis.security/">Oasis Security</a>. Oasis is building identity solutions for the agent era, reshaping one of the most important categories in enterprise security.<strong><em> </em></strong>We are joined by existing investors Cyberstarts, Sequoia, and Accel, as well as new investor Leaders Fund.</p><p>We first met the founders of Oasis, Danny Brickman and Amit Zimerman, back in 2024. The company was still early, but the founders were already operating like category leaders. They were moving with speed, recruiting with purpose, and building with the conviction that identity security needed to be rethought from the ground up.</p><p>At the time, they were pioneering the non-human identity (NHI) market. Service accounts, API keys, tokens, and machine credentials outnumber human identities by as much as 80-to-1 and organizations needed a systematic way to discover, govern, and secure them. Oasis established itself as the market leader and earned the trust of some of the largest organizations in the Global 2000.</p><p>Fast forward to today, organizations are now undergoing a seismic shift that is making the NHI market more important than ever. The catalyst is the emergence of the first truly capable AI agents, which has thrust NHI solutions from nice-to-have to must-have, critical infrastructure. In conversations with dozens of CISOs and security leaders across industries, we heard the same message repeatedly: “NHI management is now a top tier architectural priority.”</p><p>The rationale behind this shift is that NHIs ultimately control what agents can and cannot do within an organization. The problem only grows with agent adoption, as each new agent spawns additional NHIs to manage. But these agents also represent a new form of NHI that has data access, credentials, and ability to take action using corporate assets.</p><p>As companies consider their AI rollout, securing agent access is a massive concern. Legacy PAM and IGA vendors were built for a human-centric world where limiting access to permitted boundaries was sufficient. In this new world, agents are non-deterministic. Even within permitted boundaries, they need to be observed to interpret their intent behind actions and ensure they aren’t doing harm or taking actions that are misaligned with their original intent. In short, agents cannot be governed like human workers with static roles and permissions.</p><p>Oasis spent the last year building out their agent access solutions, purpose-built to solve this problem for enterprises so they can accelerate their AI adoption. Oasis can interpret an agent’s intent in real time and grant scoped, short-lived, least privileges to corporate assets as needed. The ability to map what an agent needs to do, determine the access required, and provision session-level access is a major step forward in securing this rapidly developing technology.</p><p>Identity teams can now lean on Oasis to solve two of the most acute challenges they are faced with today: managing non-human identities and controlling agent access. We believe this is just the beginning of what will become the most complete privileged access and governance platform built for the agentic era. The market opportunity represents more than $55Bn of addressable market and it is growing more than 20% a year.</p><p>Earlier this year, we were eager to reconnect with Danny and Amit to talk about the future of identity in an era of agents. What started as a quick lunch near Newark turned into time with the broader Oasis team at their Sales Kick Off in the Dominican Republic. One thing was unmistakable: Oasis is building with real force. The team they have assembled from across the identity industry is exceptionally deep, experienced, and aligned around the mission. Their energy is palpable and ambition is obvious. We spoke with several of Oasis’s Fortune 500 customers, and the feedback was about as strong as it gets. Customers consistently pointed to the team’s speed of execution, pace of product iteration, and relentless focus on customer needs.</p><p>Oasis has a very bright future ahead. We couldn’t be more excited to lead this round and deepen our partnership with the Oasis team as they define the next generation of identity security.</p><p>Written By: <a href="https://medium.com/u/1d37ba781dbe">Michael Robinson</a>, <a href="https://medium.com/u/e3cb4ee22379">Kevin Gabura</a>, <a href="https://medium.com/u/da8fff35d560">Dan Moor</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=106e0b1cfbee" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Investing in Upwind Security’s Series B, Triple Down!]]></title>
            <link>https://medium.com/@michael_robinson/investing-in-upwind-securitys-series-b-4758819e45b6?source=rss-1d37ba781dbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/4758819e45b6</guid>
            <dc:creator><![CDATA[Michael Robinson]]></dc:creator>
            <pubDate>Mon, 26 Jan 2026 18:03:27 GMT</pubDate>
            <atom:updated>2026-03-19T19:48:07.343Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xDE6z4LNpMI_N5Pr22C4cQ.jpeg" /></figure><h4>Setting the standard for what comes next in cloud security.</h4><p>We’re excited to triple down on our partnership with Upwind Security and to welcome new partners Bessemer Venture Partners and Picture Capital in the $250M Series B round. We also look forward to continued partnership with Cyberstarts, Greylock, Leaders Fund, Swish Ventures, Alta Park, TCV, Penny Jar Capital, and Cerca Partners.</p><p>From day one, founder/CEO <a href="https://medium.com/u/874ed39b3230">Amiram Shachar</a> and the Upwind team had a bold vision to build the world’s best cloud security platform. As practitioners, they had deep conviction that cloud security had to be built on a runtime-first foundation. They were right. Executing on that vision required relentless product velocity and deep customer obsession. It has clearly paid off.</p><p>Craft first invested at the seed stage before the market believed runtime belonged at the center of cloud security. Over the past few years, we’ve seen Upwind change that market perception making runtime a critical pillar of modern cloud security strategies. What gives Upwind a durable edge is using runtime context as the unifying signal across the entire CNAPP. By grounding security in what’s actually running in production, the platform dramatically improves prioritization, reduces alert fatigue, and accelerates real-world remediation.</p><p>The depth and breadth of the Upwind solution has enabled the team to win head-to-head against the most established players in the market. Upwind consistently delivers the enterprise-grade reliability, scalability, and runtime context that security leaders demand in complex, multi-cloud environments, earning the trust of some of the largest enterprises globally.</p><p>With this $250M Series B, Upwind drops into what they call “The Next Wave,” accelerating the scale-out of runtime-first cloud security to support a new era of AI-native, real-time enterprise applications worldwide. We are excited about what the team will deliver next. CNAPP lives where AI is consumed and provides the security building blocks that AI needs: Upwind is providing the security foundation that modern AI powered applications require. With AI, runtime becomes more critical today than ever before. We are starting to see the first use cases of agentic detection and response (CDR) in the cloud. AI agents can leverage runtime sensors which is critical to detecting in runtime and executing response action.</p><p>We believe Upwind is poised to not just to compete, but to dominate the next generation of cloud security platforms across application security, API security, AI security, and detection and response.</p><p><strong>Up and Upwind.</strong></p><p>Written by: <a href="https://medium.com/u/1d37ba781dbe">Michael Robinson</a> &amp; <a href="https://medium.com/u/e3cb4ee22379">Kevin Gabura</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4758819e45b6" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Our Partnership with OpenEvidence: The Intelligence Layer for Modern Medicine]]></title>
            <link>https://medium.com/@michael_robinson/our-partnership-with-openevidence-the-intelligence-layer-for-modern-medicine-e357a3560a74?source=rss-1d37ba781dbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/e357a3560a74</guid>
            <dc:creator><![CDATA[Michael Robinson]]></dc:creator>
            <pubDate>Fri, 24 Oct 2025 22:59:53 GMT</pubDate>
            <atom:updated>2025-10-28T17:05:52.950Z</atom:updated>
            <content:encoded><![CDATA[<p>By: <a href="https://medium.com/u/1d37ba781dbe">Michael Robinson</a>, <a href="https://medium.com/u/da8fff35d560">Dan Moor</a>, <a href="https://medium.com/u/69bb134c123a">Kindle Van Linge</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3y51uRxS9PGSUtgadvi_3Q.jpeg" /></figure><p>We’re thrilled to announce our investment in OpenEvidence, the AI-powered medical search platform that’s a total game-changer for the healthcare industry. Designed from the ground up for medical professionals, OpenEvidence organizes the world’s medical knowledge to make it more useful, open, accessible, and understandable.</p><p>Since launching in 2023, OpenEvidence has seen explosive adoption across the medical community, becoming the most widely used medical search engine in the U.S. Today, more than half of all U.S. physicians log in daily to make high-stakes clinical decisions. When we first heard about OpenEvidence, we asked physicians what they thought. The response was overwhelming, and it’s rare to see this level of enthusiasm for new technology in medicine:</p><p><strong><em>“It’s the best medical app I’ve ever used in my 30 year career; my team and I use it daily.”</em></strong></p><p><strong><em>“For this upcoming generation of doctors, this is a never-go-back moment.”</em></strong></p><p><strong><em>“OpenEvidence lets me think through a case with a broader differential and prompts consideration of the ‘can’t miss’ diagnoses. It’s become core to my day to day.”</em></strong></p><p>The problem OpenEvidence is solving is universal: the medical field is drowning in information. The body of medical research doubles every few months, an impossible volume for any clinician to keep up with. OpenEvidence solves this problem, bringing real-time, peer-reviewed insights directly to clinicians at the point of care.</p><p>In an industry where trust is everything,<em> </em>physicians need answers grounded in science and delivered with speed and context. Unlike general-purpose large language models, OpenEvidence is purpose-built for medicine and trained exclusively on high-quality, peer-reviewed sources. Its partnerships with The New England Journal of Medicine and JAMA provide a foundation of credibility that sets a new standard for accuracy in clinical AI. OpenEvidence has collected data on more than 100 million clinical consultations, and earlier this year became the first AI platform in history to score 100% on the US Medical Licensing Exam. The company now powers more than 15 million clinical consultations every month. To put that in perspective, in 2025, roughly one in three Americans will be treated by a doctor using OpenEvidence.</p><p>Behind this momentum are co-founders Daniel Nadler (CEO) and Zach Ziegler (CTO), and a world-class team of PhD researchers, engineers, doctors, and operators. As a two-time founder, Daniel has spent his career building tools that make complex information accessible and actionable. Before OpenEvidence, he founded Kensho, the pioneering AI analytics platform acquired by S&amp;P Global.</p><p>With a stellar reputation in the medical community, it’s no surprise that OpenEvidence is attracting exceptional technical talent and experienced healthcare veterans. The team has completely reimagined how technology is delivered to healthcare providers and aligned the business directly with the interests of clinicians, building loyalty from the ground up.</p><p>We couldn’t be more excited about the mission and to be partnering with Daniel, Zach, and the entire OpenEvidence team as they build the intelligence layer for modern medicine.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e357a3560a74" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why We Invested in Starbridge]]></title>
            <link>https://medium.com/@michael_robinson/why-we-invested-in-starbridge-6f0ad3f32a6b?source=rss-1d37ba781dbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/6f0ad3f32a6b</guid>
            <dc:creator><![CDATA[Michael Robinson]]></dc:creator>
            <pubDate>Wed, 22 Oct 2025 21:18:21 GMT</pubDate>
            <atom:updated>2025-11-28T18:34:58.500Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Powering the future of SLED sales and servicing the American public</em></p><p>By: <a href="https://medium.com/u/1d37ba781dbe">Michael Robinson</a>, <a href="https://medium.com/u/3b6846dd328">J Zac Stein</a>, <a href="https://medium.com/u/d7706d1a9fd9">Zao Chen</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xU51uTpwUU0Q9paKTQ2tfA.png" /></figure><p>We’re excited to lead the $42M Series A round in Starbridge.ai, the AI-powered intelligence platform for companies selling into state, local government, and educational (SLED) institutions. <a href="https://x.com/ZacBookman">Zac Bookman</a> (OpenGov) and <a href="https://x.com/mwseibel">Michael Seibel</a> (Y Combinator), as well as existing investors Owl Ventures, Autotech Ventures, and Commonweal Ventures, are joining us in this round to support Starbridge on their journey to help companies service the American public sector.</p><p>When Justin Wenig, Founder and CEO, was previously building Coursedog, he experienced the frustration of selling to SLED firsthand. From combing through hundreds of pages of meeting minutes to manually requesting data through FOIA requests, Justin and his team were spending hours just to figure out who to reach out to. With AI and LLMs becoming increasingly powerful in distilling down data sets, Justin decided to build the product he wished he had at Coursedog, which ultimately led to the genesis of Starbridge. We’ve known Justin for a while and he’s an absolute force. From our first meeting, we knew Justin had the perfect combination of domain expertise, product intuition, and founder grit to build something truly transformative. Having lived the pain point and built a successful company before, Justin brings rare insight and intensity to redefine how companies access, interpret, and act on public-sector data.</p><p>The annual US SLED budget is massive, accounting for ~10% of US GDP at nearly $2 trillion and spanning across tens of thousands of entities from states, municipalities, and counties. SLED institutions often contract private sector organizations to serve the public in areas such as IT, education, and healthcare. However, they are met with a complex web of data, procurement processes, cooperative purchase agreements, and manual workflows such as FOIA requests to win SLED contracts. This has made it challenging for even the best vendors and has limited healthy competition to service the American public. Starbridge solves this problem by empowering every company with the ability to discover and win these opportunities.</p><p>Starbridge’s comprehensive platform for SLED procurement uses AI agents to gather data from public records (e.g. RFPs, meeting minutes) and automates traditionally manual workflows like FOIA requests. This allows customers to access highly tailored intelligence that would otherwise be expensive. Simultaneously, it creates a data flywheel with more users submitting FOIA requests, and a growing contract dataset. Using LLMs, Starbridge distills this growing dataset into actionable insights that save sales teams hours of research time. Since its launch in early 2025, Starbridge has grown rapidly to 150+ customers. When we spoke with customers, the theme that was echoed was “exceptional ROI.” Starbridge reduces hours of manual work, all while surfacing signals that enable users to significantly accelerate their ability to find and win high-value contracts.</p><p>Starbridge is quickly becoming an essential part of companies’ SLED sales strategy and we couldn’t be more thrilled to partner with Justin and the Starbridge team.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6f0ad3f32a6b" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>