How to Support Developers With AI

Tips for AI-Assisted software development: Treat AI like a pair programmer, not a code vending machine. The most useful mental model for AI-assisted engineering is collaboration. When you see AI as a pair, you stay in control. You guide, review, challenge, and refine. The quality goes up because you’re thinking together, not outsourcing your judgment. It’s the same discipline you’d apply with a human partner. One drives, one navigates. As the navigator, question the code and challenge the assumptions. Do this in practice: - Be the driver. Let AI write code, you focus on architecture, edge cases, and security. - Keep it conversational. Explain your intent, then iterate. Treat prompts as dialogue, not commands. - Ask it to explain its own code. If you can’t follow the explanation, don’t merge the code. - Trust, but verify. Check APIs, versions, and performance assumptions. Run the tests every time. - Use it as a rubber duck. Explaining the problem often reveals the solution. - Challenge suggestions that feel off. Probe edge cases and trade-offs. - Switch who’s driving. Stay engaged so you keep ownership of the code. - Step away when needed. Blind acceptance is a smell, even with AI. Manage the context to stay relevant and focused. - Think of AI as a brilliant, fast and naive developer. Huge range, zero business context, and no common sense about your business. Your job is to pair well.

Most AI coders (Cursor, Claude Code, etc.) still skip the simplest path to reliable software: make the model fail first. Test-driven development turns an LLM into a self-correcting coder. Here’s the cycle I use with Claude (works for Gemini or o3 too): (1) Write failing tests – “generate unit tests for foo.py covering logged-out users; don’t touch implementation.” (2) Confirm the red bar – run the suite, watch it fail, commit the tests. (3) Iterate to green – instruct the coding model to “update foo.py until all tests pass. Tests stay frozen!” The AI agent then writes, runs, tweaks, and repeats. (4) Verify + commit – once the suite is green, push the code and open a PR with context-rich commit messages. Why this works: -> Tests act as a concrete target, slashing hallucinations -> Iterative feedback lets the coding agent self-correct instead of over-fitting a one-shot response -> You finish with executable specs, cleaner diffs, and auditable history I’ve cut debugging time in half since adopting this loop. If you’re agentic-coding without TDD, you’re leaving reliability and velocity on the table. This and a dozen more tips for developers building with AI in my latest AI Tidbits post https://lnkd.in/gTydCV9b

Is AI automating away coding jobs? New research from Anthropic analyzed 500,000 coding conversations with AI and found patterns that every developer should consider: When developers use specialized AI coding tools: - 79% of interactions involve automation rather than augmentation - UI/UX development ranks among the top use cases - Startups adopt AI coding tools at 2.5x the rate of enterprises - Web development languages dominate:          JavaScript/TypeScript: 31%          HTML/CSS: 28% What does this mean for your career? Three strategic pivots to consider: 1. Shift from writing code to "AI orchestration"     If you're spending most of your time on routine front-end tasks, now's the time to develop skills in prompt engineering, code review, and AI-assisted architecture. The developers who thrive will be those who can effectively direct AI tools to implement their vision. 2. Double down on backend complexity     The data shows less AI automation in complex backend systems. Consider specializing in areas that require deeper system knowledge like distributed systems, security, or performance optimization—domains where context and specialized knowledge still give humans the edge. 3. Position yourself at the startup-enterprise bridge     With startups adopting AI coding tools faster than enterprises, there's a growing opportunity for developers who can bring AI-accelerated development practices into traditional companies. Could you be the champion who helps your organization close this gap? How to prepare: - Learn prompt engineering for code generation - Build a personal workflow that combines your expertise with AI assistance - Start tracking which of your tasks AI handles well vs. where you still outperform it - Experiment with specialized AI coding tools now, even if your company hasn't adopted them - Focus your learning on architectural thinking rather than syntax mastery The developer role isn't disappearing—it's evolving. Those who adapt their skillset to complement AI rather than compete with it will find incredible new opportunities. Have you started integrating AI tools into your development workflow? What's working? What still requires the human touch?

AI Tools That Genuinely Boosted My Productivity as a Software Engineer After trying dozens of AI tools over the past few months, I’ve narrowed down the list to a few that truly made a difference in my workflow. These tools have helped me code faster, understand complex systems better, and reduce repetitive tasks. Here are the top ones that stuck with me: 1. GitHub Copilot – For coding assistance  Suggests lines, functions, even entire files.  I use it daily in VS Code to autocomplete logic, generate test cases, and eliminate boilerplate code. 2. CodeWhisperer by AWS – Secure code generation  An AWS-native alternative to Copilot, focused on security and privacy.  It’s extremely helpful when integrating AWS SDKs and working on backend services. 3. Phind – Dev-specific AI search  This replaced Google for me when it comes to technical questions.  Phind gives concise, accurate answers for framework issues, error debugging, and best practices. 4. Tabnine – Secure and private code completion  Great when you’re working with sensitive or proprietary code.  Runs on-prem and supports a wide range of languages and IDEs. 5. Codeium – Lightweight code autocomplete  A fast and free alternative to Copilot.  I use it for side projects, and it performs well with multiple languages and frameworks. 6. Cody by Sourcegraph – Chat with your codebase  Lets me ask questions like “What does this function do?” or “Where is this used?”  It’s a major help when exploring large or legacy codebases. These tools helped me: Debug faster Refactor smarter Document better Ship cleaner code If you're a developer and haven’t explored these yet, start with GitHub Copilot or Phind. They’re game changers. What AI tools are you currently using in your dev stack? Always open to trying more. Follow Abhay Singh for more such reads.

🔍𝗔𝗜 𝗶𝗻 𝗦𝗼𝗳𝘁𝘄𝗮𝗿𝗲 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 - 𝗦𝗸𝗶𝗹𝗹 𝗔𝗰𝗰𝗲𝗹𝗲𝗿𝗮𝘁𝗼𝗿 𝗼𝗿 𝗦𝗸𝗶𝗹𝗹 𝗖𝗿𝘂𝘁𝗰𝗵? After 18 years in the software industry, working closely with many engineers (especially junior and mid-level engineers), I’ve always been cautiously optimistic about AI. But I’ve always had a concern: 👉 If AI is used carelessly, it may reduce real learning instead of accelerating it. Today, I found strong evidence supporting that intuition - not just from experience, but from rigorous research by Anthropic: 📌 AI Assistance Can Impair Learning (Coding Skills Study) https://lnkd.in/gygkb_CY 🧠 𝗞𝗲𝘆 𝗙𝗶𝗻𝗱𝗶𝗻𝗴𝘀 Anthropic studied developers solving a coding task while learning a new library. Here’s what they found: -> Developers using AI finished slightly faster -> But their understanding was significantly weaker 📉 In a follow-up mastery quiz, AI-assisted developers scored ~17% lower than those who coded without AI. Even more interesting: 🔍The biggest skill gap was in debugging + comprehension - the exact skills required to build robust, maintainable software and to understand why something works (or breaks). 💡 𝗧𝗵𝗲 𝗠𝗼𝘀𝘁 𝗜𝗺𝗽𝗼𝗿𝘁𝗮𝗻𝘁 𝗜𝗻𝘀𝗶𝗴𝗵𝘁 Not all AI usage harms learning. The study showed a clear difference between two groups: Passive AI Users: - copy/paste code - accept suggestions blindly - treat AI as an auto-complete machine Active AI Learners: - ask “why?” - request explanations - explore alternatives - validate with their own reasoning And guess what? Active AI learners performed much better. 🚀 𝗪𝗵𝗮𝘁 𝗧𝗵𝗶𝘀 𝗠𝗲𝗮𝗻𝘀 𝗳𝗼𝗿 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗲𝗿𝘀 AI is not a shortcut to expertise. It can make you faster, but speed without understanding is risky. To thrive in the AI era: 1. Strengthen fundamentals (CS basics, design, debugging) 2. Use AI intentionally - like a mentor, not an auto-complete tool 3. Focus on deep understanding and engineering judgment 🎯 👥 𝗪𝗵𝗮𝘁 𝗧𝗵𝗶𝘀 𝗠𝗲𝗮𝗻𝘀 𝗙𝗼𝗿 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 𝗠𝗮𝗻𝗮𝗴𝗲𝗿𝘀 Don’t just encourage AI usage. Encourage structured learning with AI. Build cultures where engineers: - question AI outputs - explain decisions - learn deliberately - debug deeply instead of patching quickly Because the future belongs to engineers who can do both: 🚀 Move fast with AI 🧠 Think deeply without it #SoftwareEngineering #AI #DeveloperSkills #Leadership #Learning

We've entered a phase where AI can be involved in more than just code generation. But new research shows that adoption is not about what AI can do... it's about what developers want it to do—and why. A study of 860 developers at Microsoft reveals that task characteristics—not just capability—drive AI adoption. 🔝 High-value + high-demand work → developers seek AI help 🤳 Identity-aligned work → developers resist giving up control 📋 High-accountability work → developers use AI but insist on oversight The research showed three "zones" for AI use cases: ⚒️ Build/Improve: Core technical work (coding, testing, debugging, code review) has strong AI demand, but developers want augmentation, not automation. They'll use AI to handle boilerplate and reduce cognitive load—but decision control stays human. 📉 De-prioritize: People & strategic work (mentoring, stakeholder communication, system design) has low AI appetite. These tasks require empathy, relationships, and contextual judgment that AI shouldn't own. 🌟 Opportunity gaps: Ops & coordination (DevOps, documentation, infrastructure monitoring) see high demand... but low adoption. This is because devs need to see reliability, privacy/security, and transparency before trusting it further. This study is a reminder that with any tool, AI has boundaries of value. To get the most value, first map where AI fits in your team's actual work. Then deploy it to crush toil, use it to augment technical work, and keep it peripheral in strategy and relationships. The goal is not to automate developers, but to clear space for them to do work that matters so we build better products.

I’m staring at a screen full of error messages. It’s 2am. The deployment failed again. A junior developer messages me: “I’ve spent 12 hours on this bug. No progress.” This was my reality two years ago. We’d just adopted an AI code assistant. The hype promised “10x productivity.” The reality? Frustration. Mistrust. Burnout. Here’s what no one told me: AI doesn’t fix broken workflows. It amplifies what’s already there. Our first mistake? We rolled out the tool without asking developers what they needed. We solved the wrong problem. The real issue wasn’t writing code faster. It was fixing bugs. Waiting for tests. Navigating legacy systems. We’d bought a scalpel for a sledgehammer job. Then came the turning point. A senior engineer asked: “What if we use this for code reviews instead?” We shifted focus. Trained the AI to catch common errors before pull requests. Integrated it into CI/CD pipelines. Measured impact weekly. Within 3 months: Review cycles dropped from 3 days to 11 hours. Merge conflicts fell by 60%. Developers started asking, “How did we miss this?” One team lead joked: “I’m terrified of going back to manual reviews now.” The lesson? AI isn’t magic. It’s a mirror. If you point it at the wrong problems, it magnifies waste. If you listen to your team first, it becomes a force multiplier. The LeadDev survey makes sense now. 6% see real gains because few ask developers: Where does it hurt? What would save you 3 hours today? What tool would make your job… less of a slog? Last week, a new developer asked: “Do you think AI will replace us?” I thought back to that 2am debugging session. “No,” I said. “But the developer who uses it to fix bugs at 10pm on a Friday? They’ll replace the one who doesn’t.” Because AI isn’t here to take over. It’s here to give engineers back the one thing we always run out of: Time. Time to build. Time to create. Time to solve the problems that matter. The rest? Let the machines handle it.

Developers don’t need another tool. They need AI that works where they already live—inside their IDE, CLI, and team workflows. The rise of tools like Cursor proves a critical point: the best AI is somewhat invisible. Cursor is a fork of VS Code. No new tabs, no context switching, no forcing teams to adopt foreign workflows. It meets developers exactly where they are—enhancing, not interrupting, their flow. This is the blueprint for AI adoption in tech: ✅ Seamless integration: Tools that feel like a natural part of the existing toolkit (like Cursor in VS Code) avoid the “adoption tax.” ✅ Context is king: AI that leverages your local codebase, open files, and even unresolved Git conflicts becomes truly useful. ✅ Trust through utility: When AI assists without demanding attention (e.g., inline code suggestions, quiet error detection), it earns its place as a teammate. The lesson? AI succeeds when it’s frictionless. Developers won’t bend their workflows for flashy tech—but they’ll embrace AI that respects their process. Tools like Cursor aren’t just “nice-to-have”—they’re becoming core infrastructure because they amplify expertise instead of complicating it.

I shipped 100,000 lines of high-quality code in 2 weeks using AI coding agents. But here's what nobody talks about: we're deploying AI coding tools without the infrastructure they need to actually work. When we onboard a developer, we give them documentation, coding standards, proven workflows, and collaboration tools. When we "deploy" a coding agent, we give them nothing and ask them to spend time changing their behavior and workflows on top of actively shipping code. So I compiled what I'm calling AI Coding Agent Infrastructure or the missing support layer: • Skills with mandatory skill checking that makes it structurally impossible for agents to rationalize away test-driven development (TDD) or skip proven workflows (Credits: Superpowers Framework by Jesse Vincent, Anthropic Skills, custom prompt-engineer skill based on Anthropic’s prompt engineering overview). • 114+ specialized sub-agents that work in parallel (up to 50 at once) like Backend Developer + WebSocket Engineer + Database Optimizer running simultaneously, not one generalist bottleneck (Credits: https://lnkd.in/dgfrstVq) • Ralph method for overnight autonomous development (Credits: Geoffrey Huntley, repomirror project https://lnkd.in/dXzAqDGc) This helped drive my coding agent output from inconsistent to 80% of the way there, enabling me to build at a scale like never before. Setup for this workflow takes you 5 minutes. A single prompt installs everything across any AI coding tool (Cursor, Windsurf, GitHub Copilot, Claude Code). I'm open sourcing the complete infrastructure and my workflow instructions today. We need better developer experiences than being told to "use AI tools" or manually put all of these pieces together without the support layer to make them actually work. PRs are welcome, whether you're building custom skills, creating domain-specific sub-agents, or finding better patterns. Link to repo: https://lnkd.in/dfm4NAmh Full breakdown of workflow here: https://lnkd.in/dr9c-UX3 What patterns have you found make the biggest difference in your coding agent productivity?

"Vibe Coding !== Low Quality Work: a guide to responsible AI-assisted dev" ✍️ My latest free article: https://lnkd.in/gjMdjMWV The allure of "vibe coding" – using AI to "move faster and break even more things" – is strong. AI-assisted development is undeniably transformative, lowering barriers and boosting productivity. But speed without quality is a dangerous trap. Relying uncritically on AI-generated code can lead to brittle "house of cards" systems, amplify tech debt exponentially, and introduce subtle security flaws. Volume ≠ Quality. A helpful mental model I discuss (excellently illustrated by Forrest Brazeal) is treating AI like a "very eager junior developer." It needs guidance, review, and refinement from experienced hands. You wouldn't let a junior ship unreviewed code, right? So how do we harness AI's power responsibly? I've outlined a field guide with practical rules: ✅ Always review: Treat AI output like a PR from a new hire. ✅ Refactor & test: Inject engineering wisdom – clean up, handle edge cases, test thoroughly. ✅ Maintain standards: Ensure AI code meets your team's style, architecture, and quality bar. ✅ Human-led design: Use AI for implementation grunt work, not fundamental architecture decisions. The goal isn't to reject vibe coding, but to integrate it with discipline. Let's use AI to augment our craft, pairing machine speed with human judgment. #softwareengineering #programming #ai