<?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 Adelanke Adeleye on Medium]]></title>
        <description><![CDATA[Stories by Adelanke Adeleye on Medium]]></description>
        <link>https://medium.com/@adelanke?source=rss-b8c120e55c72------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*SoD0ZN46gA0GIHI1Ucwmiw.jpeg</url>
            <title>Stories by Adelanke Adeleye on Medium</title>
            <link>https://medium.com/@adelanke?source=rss-b8c120e55c72------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 23:01:08 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@adelanke/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[AI Isn’t Bad at Coding. We’re Just Not Giving It Enough Context.]]></title>
            <link>https://medium.com/@adelanke/ai-isnt-bad-at-coding-we-re-just-not-giving-it-enough-context-6784645c76ae?source=rss-b8c120e55c72------2</link>
            <guid isPermaLink="false">https://medium.com/p/6784645c76ae</guid>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[prompt-engineering]]></category>
            <category><![CDATA[product-development]]></category>
            <category><![CDATA[ai-assisted-development]]></category>
            <category><![CDATA[context-engineering]]></category>
            <dc:creator><![CDATA[Adelanke Adeleye]]></dc:creator>
            <pubDate>Sat, 18 Jul 2026 13:32:49 GMT</pubDate>
            <atom:updated>2026-07-18T14:35:07.530Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OOvvYY6f4KWV-E-l1nCeoQ.png" /><figcaption>Why I stopped chasing better prompts and started building better context and how it completely changed my AI development workflow.</figcaption></figure><p>A few weeks ago, I noticed something that kept happening no matter which AI coding tool I used.</p><p>I’d ask it to build a landing page.</p><p>It looked great.</p><p>Then I’d ask for a dashboard.</p><p>Later I’d add authentication, settings, and a few more features.</p><p>At first, everything seemed fine. But after a few sessions, the project started to feel inconsistent. Components looked different, naming conventions had changed, business logic had found its way into UI components, and the codebase no longer felt like it belonged to one team.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8gjnI0ru3BaBBHNNajfwGg.png" /></figure><p>I assumed the problem was the AI.</p><p>I switched between different tools, rewrote prompts, and experimented with different workflows, hoping one of them would solve the problem.</p><p>It never really did.</p><p>Then I came across Context Engineering, and it completely changed how I think about building software with AI.</p><h3>The AI Wasn’t Failing. It Was Guessing.</h3><p>The realization that changed everything was surprisingly simple.</p><p>The AI never had a company handbook.</p><p>It had never seen a Product Requirements Document.</p><p>It didn’t know my coding standards.</p><p>It didn’t understand the architecture.</p><p>It had no idea how I wanted folders organized or how different parts of the application were supposed to work together.</p><p>Every new conversation was like hiring a talented engineer, giving them a single task, and expecting them to understand an entire company without any onboarding.</p><p>Of course they would make assumptions.</p><p>AI is no different.</p><p>Whenever information is missing, it fills in the gaps using what it already knows.</p><p>Sometimes those assumptions are exactly what you wanted.</p><p>Sometimes they’re completely different.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*V-i_yD1MVfa8VrziVrpfbg.png" /></figure><p>The more I thought about it, the more I realized I wasn’t giving the AI enough context to make consistent decisions.</p><h3>Prompt Engineering Helps Conversations. Context Engineering Helps Projects.</h3><p>Prompt engineering is still important.</p><p>A well written prompt gives the AI better instructions for the task you’re asking it to complete.</p><p>But prompts only solve the conversation you’re having right now.</p><p>Context Engineering solves the project.</p><p>Instead of relying on one carefully written prompt after another, you’re creating a shared source of truth that every future conversation can build on.</p><p>That was the shift that changed my workflow.</p><p>Instead of asking, “How can I write a better prompt?”</p><p>I started asking, “What would a new engineer need to know before writing their first line of code?”</p><p>The answer turned into a simple checklist.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*be4F0A2r_NDdLahw1EPT8Q.png" /></figure><p>A Product Requirements Document that clearly explains what we’re building and what success looks like.</p><p>An AGENTS.md file that acts as the project’s handbook, defining conventions, standards, and expectations.</p><p>Specialized rule files so each AI agent understands its responsibilities instead of trying to do everything.</p><p>A predictable project structure so every file has a clear home.</p><p>None of these documents are complicated on their own.</p><p>Together, they give the AI enough context to stop making random assumptions and start making more consistent decisions.</p><h3>Putting the Idea Into Practice</h3><p>Learning the theory wasn’t enough.</p><p>I wanted to see if it actually worked.</p><p>So I created <strong>RIV</strong>, an example project that became my testing ground for Context Engineering.</p><p>RIV isn’t a production product.</p><p>It’s a reference project that I use to experiment with workflows, test different approaches, and understand what actually improves AI assisted development.</p><p>Using one consistent project made it much easier to compare ideas, refine my process, and document everything I learned.</p><p>Over time, I realized I wasn’t just experimenting anymore.</p><p>I was building a repeatable workflow.</p><p>One that worked whether I was using Cursor, Claude Code, Windsurf, VS Code, or any other AI coding tool.</p><p>That was another lesson I didn’t expect.</p><p>The tool wasn’t the most important part.</p><p>The context was.</p><p>You can switch editors tomorrow.</p><p>You can switch models next month.</p><p>Good context still works.</p><p>The tool is just where the work happens.</p><p>The context is what guides the work.</p><h3>Why I Decided to Write a Guide</h3><p>As I kept refining this workflow, my notes slowly turned into something much bigger.</p><p>I had a complete PRD.</p><p>An AGENTS.md file.</p><p>Agent rule files.</p><p>A project structure.</p><p>Documentation explaining why each piece exists and how they work together.</p><p>Rather than keeping those notes to myself, I decided to turn them into a practical guide.</p><p>Not because I’ve mastered Context Engineering.</p><p>I’m still learning every time I build something new.</p><p>I wrote it because it’s the guide I wish I’d had when I started building with AI.</p><p>Everything in the guide is based on the workflow I developed while using RIV as a reference project.</p><p>It includes the actual documents, the reasoning behind them, and practical examples that readers can adapt to their own projects instead of starting from scratch.</p><h3>One Lesson Keeps Coming Back</h3><p>If there’s one thing this journey has taught me, it’s this.</p><p>Better prompts improve conversations.</p><p>Better context improves products.</p><p>The more I build with AI, the more I believe the biggest challenge isn’t choosing the right model.</p><p>It’s giving that model enough context to make good decisions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YtP0vvSb-icNL-LVKp59hg.png" /></figure><p>I’m still learning.</p><p>I’m still experimenting.</p><p>But this one idea has completely changed the way I build software, and I think it’s going to become one of the most important skills for anyone building with AI.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6784645c76ae" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[DESIGNER VS CLIENT/STAKEHOLDER (Getting approvals & managing relationships)]]></title>
            <link>https://medium.com/adobexd-nigeria-community/designer-vs-client-stakeholder-getting-approvals-managing-relationships-4c234f27d060?source=rss-b8c120e55c72------2</link>
            <guid isPermaLink="false">https://medium.com/p/4c234f27d060</guid>
            <category><![CDATA[design]]></category>
            <category><![CDATA[nigeria]]></category>
            <category><![CDATA[adobe-xd]]></category>
            <category><![CDATA[ama]]></category>
            <category><![CDATA[client-relationship]]></category>
            <dc:creator><![CDATA[Adelanke Adeleye]]></dc:creator>
            <pubDate>Tue, 12 Feb 2019 16:01:02 GMT</pubDate>
            <atom:updated>2019-02-12T16:01:02.402Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*lRI8OYE4fkgaUkveyP9IAw@2x.jpeg" /></figure><h3>Designer VS Client/Stakeholder (Getting approvals &amp; managing relationships)</h3><h4>AdobeXD Nigeria Community AMA With Leslie Williams</h4><p>Leslie Williams is the Creative Director at TeamApt. He is a product designer for mobile/web/games, Visual Identity designer for brands and he dabbles into motion graphics whenever he can. He is also the co-founder of the JobCaster network (Design.Jobcaster &amp; Dev.Jobcaster ) which helps connect talents with the people who want to hire them. When he is not working, he curates creative work on <a href="http://instagram.com/layrz_ng">Instagram</a>, shares how he works on his <a href="https://www.youtube.com/c/PlayBookTV">youtube channel</a>, reading and basically doing anything that catches his fancy.</p><p><strong>Leslie Williams </strong>— <em>“You’ve been given a brief from the client, outlining what he wants. He/she got your contacts from a friend of a friend of a friend and everyone is hyped you that you do really good work for them. So for you, this is a walk in the park. Then you set about to working on the brief using your usual creative pipeline: Faff for a week under the guise of “discovery &amp; research”, take another week to nail down a concept and as the deadline is looming, you enter overdrive to deliver what you believe to be the best fit solution for the client’s needs.</em></p><p>Then when you show the client, he/she takes one look at this and says one of the following:</p><p><em>- I don’t like it</em></p><p><em>- It needs to pop more</em></p><p><em>- Why doesn’t it look like XYZ’s one?</em></p><p>No designer likes that. Ever.</p><p>As designers, we often fantasize about designing just once and the client being blown away and accepts the deliverables fully, no questions asked. In reality, this hardly ever happens.</p><p>This situation is usually true for freelance/entrepreneurial designers as well as designers working 9–5 in this case, their clients are both internal (product/project managers are the stakeholders) or external i.e the brands/businesses they work with</p><p>In the course of this discussion, I will approach this from two angles;</p><p>1. How to ensure that the client/stakeholder is onboard with your work so as to minimize friction/drama</p><p>2. How to manage the Clients/Stakeholders when they aren’t satisfied with your output</p><p>So, be it your boss, your PMs (both product &amp; project) and even external clients that you engage with — They’re all human. They want the best work from you and they want the best solution that meets their diverse needs. So it helps to have an understanding of the kinds of people you’ll be working with.</p><p><strong>Understand the frog before you eat it.</strong></p><p>Take out time to study your client/stakeholder. Usually, you get to do this from your project meeting.</p><p>Listen to their conversations: What words do they repeat the most? How do they carry themselves, How do they see themselves? Are they very articulate in communicating their ideas or do they just gloss over everything and expect you to fill in the gaps? Are they savvy about the subject matter or are they “winging it”?</p><p>This helps you evaluate their character by stepping into their mindset. It’s also advantageous as the project progresses down the line, as you’ll streamline your output more to their expectations/taste.</p><p><strong>Make your process as transparent as possible and walk them through every step</strong></p><p>This part lies with you. It’s important that you’ve got every step in your creative pipeline clearly defined and that you can explain these steps to the client/stakeholder such that they can understand and plan their expectations around it.</p><p>Some of them are not properly educated enough to know that design is a process from start to finish and even beyond the final deliverable. Imagine going for a review meeting to showcase wireframes and user-flows and the stakeholders think the rough sketchy lines are the actual finished product.</p><p>You’d probably spend another 20 minutes trying to explain this to them before they’re convinced and even then, it might cost you. So it’s important that you educate them about the design process early, preferably before you accept the project. While you’re defining the scope of the project with the client</p><p><strong>Ask questions, document the responses and share with the client to ensure everyone is on the same page.</strong></p><p>So usually, what I do is I open excel or google spreadsheet (don’t get triggered, spreadsheet lives matters too) and I try to do a breakdown of the scope into features &amp; tasks after which I’ll share with the client. More often than not, there’d be something I missed out or didn’t get clear enough and it’d be up to the client to provide clarity and context on those things.</p><p>Once everyone is in agreement, I use the spreadsheet as my task-list, such that as soon as I complete a feature or a milestone, I check it as “awaiting client review”. Once the client has reviewed and is satisfied with it, he checks it as approved. Again, this ensures everyone is on the same page. Plus, it helps curtail the client/SH’s excesses.</p><p>For freelancers, it goes without saying that you’ve got to have a ready defense for when clients start becoming excessive. Remember: rework means more time, effort and resources from your end *without pay*</p><p><strong>Feedback, deadlines &amp; milestone set.</strong></p><p>Setting a deadline is very tricky. Don’t fall into the trap of trying to impress the client with speed. You’ll see yourself sacrificing quality for that mistake. What I usually do is evaluate the time I know I can execute, then pad it with 1–2 weeks.</p><p><strong>Upon delivery, walk them through the output again and run it against your process pipeline.</strong></p><p>This part is the tricky part, so you’ve got to be extra careful here. Usually at this point, one small mistake and you suddenly find yourself in rework hell, so be sure to review your work (preferably among your peers to get professional feedback) Once everything is done and the client checks that everything conforms, then you can sit back, relax, take that freshly brewed palm-wine and wait for your account balance to update its status.</p><p><em>So what now happens when you’ve gotten to the point when the client is dissatisfied with your output?</em></p><p><strong>Don’t panic. Don’t throw a tantrum. Calm down.</strong></p><p>Sometimes as creatives we get overly attached to our creations. However, we’ve got to exhibit empathy: the ability to place ourselves in our clients/SH’s shoes and see what we’ve done from another light. Be open-minded and don’t take feedback as a personal assault (except those ones that are especially rude and are waiting for you to send them to hell on a first class ticket)</p><p><strong>Resist the urge to just dive into redesigning, engage them and ask them what is wrong.</strong></p><p>Sometimes, what they might have issues with might just be something they don’t understand so you’ve got to be on-hand to explain it to them and sway them as well. If it’s a critical issue, be gracious enough to step up, accept responsibility, ask more questions to help you better understand it and refine it to a workable solution.</p><p><strong>The key theme in all I’ve said is *Communication*.</strong></p><p><em>Always work closely with your clients and stakeholders.</em></p><p><strong>Questions from the community</strong></p><p><strong>Question — </strong>How can we get more companies educated about design and its benefits</p><p><strong>Leslie Williams — </strong>I’m not sure there’s a straightforward answer to that question. In the case of In-house designers, the onus falls on the creatives to continually educate the stakeholders to “buy-in” into a design-centered culture. This is usually hard when the people you report to aren’t progressives. Even worse, in Nigeria, where we have this “rush” culture where in order to meet with timelines, you’ve got to cut corners and skips steps. But I think if you’ve got one person on your side who has seen the benefits, it’d go a long way for advocating for the design culture company-wide.</p><p>For Freelancers, you’ve got to sit with your clients and walk them through the process, explaining each step. They may not buy-in immediately until they see evident value in the process.</p><p><strong>Question — </strong>How do you handle those clients that Don’t even know what they want exactly, and give you sample products that seem to contrast each other as inspiration?</p><p><strong>Leslie Williams </strong>— Well, it can go one of two ways. You can sit with the client and drill down exactly what he/she wants. It’d be a grueling process but both of you will be better for it.</p><p>Walk away. Decline graciously, help the client by recommending another creative whom you think would be a good fit then wish him/her the best. Just don’t burn bridges.</p><p><strong>Question — </strong>What do you think about the state of design and design education in Nigeria. What non-design skills should a designer have to build influence in the company or with client</p><p><strong>Leslie Williams — </strong>I don’t want to speak on the state of design in Nigeria because it’s something I’m very passionate about and this could degenerate into a full-blown rant. However, I think things are improving. Individuals (mostly creatives) are going out of their way to help empower fellow designers and others looking to become designers by organizing workshops and seminars. I look forward to a time when the average Nigerian is design conscious and I believe we can make it happen in our time.</p><p>Regarding what non-design skills to have to build influence and clout, basically, I’d say</p><ul><li>Learn to communicate effectively</li><li>Learn to negotiate effectively such that you get results that favor you, even in the tiniest degree. Small victories often stack up to bigger ones.</li></ul><p><strong>Question — </strong>As an in-house designer, how do you work with a CEO whose desire is to be involved at every step of the project, can this be seen more like a curse than a blessing. Like every time during evaluation, he comes up with new ideas and features that weren’t part of the project scope and that indirectly affect your timeline.</p><p><strong>Leslie Williams — </strong>I think, in as much as you can’t change their behavior immediately, the best you can do is to win their trust in your competency over time. I had this same issue with my current boss. However, as we worked together, he began to trust my decisions and backed off to the point where he doesn’t even check on me. Once he knows I’m in charge, he needs not to lose any sleep.</p><p>Give a clap or two if you enjoyed this story. Like this type of content? do follow us on <a href="http://lnkd.in/dmrixnt">Twitter</a> &amp; <a href="http://lnkd.in/d4saZ8T">Instagram</a>, you can also be a part of the AdobeXD Nigeria Community on <a href="http://bit.ly/AdobeXDNigeria">Slack</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4c234f27d060" width="1" height="1" alt=""><hr><p><a href="https://medium.com/adobexd-nigeria-community/designer-vs-client-stakeholder-getting-approvals-managing-relationships-4c234f27d060">DESIGNER VS CLIENT/STAKEHOLDER (Getting approvals &amp; managing relationships)</a> was originally published in <a href="https://medium.com/adobexd-nigeria-community">AdobeXD Nigeria Community</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>