Developer Tea

When Good Thinking Becomes Overthinking: Discover why the pursuit of perfect analysis often undermines good decision-making. Loading every caveat, every exception, and every alternative into your working memory doesn't produce better outcomes — it produces paralysis. Heuristics as a Feature, Not a Bug: Your brain is an efficiency machine that creates shortcuts — cached concepts, stored routines, snap judgments. These heuristics are always incomplete, but they let you move through complex problems quickly. The opportunity is to deliberately choose which heuristics to exploit. "All Models Are Wrong, Some Models Are Useful": Useful illusions don't need to be perfectly true. They need to be true enough that acting on them produces better outcomes than endlessly debating their accuracy. Useful Illusion: Coding by Hand Is Going Away: Whether or not this is literally true in every case, the engineer who acts as if it is will invest in agentic workflows, LLMs, and new tooling — while the engineer who picks the argument apart risks being labeled a skeptic and falling behind. Useful Illusion: Hard Work Pays Off: You can poke holes in this all day — wrong direction, burnout, culture-dependent — but people who follow this heuristic tend to build reputations as reliable and capable. Few of us want to be known for the opposite. Useful Illusion: As Long As I'm Learning, I'm Growing: Learning becomes less directly correlated with career advancement over time, but continuing to act on this belief keeps you flexible, curious, and in a growth mindset. More Useful Illusions for Your List: Clean code is better. Always think about the user's experience. Go with the tool you know. Volume of delivered work correlates with career success — especially during performance review season. The Key Insight: You don't have to believe any of these things literally. You're exploiting your own heuristic system to drive efficient action and avoid wasting time on low-utility debates. The result is a more decisive, action-oriented version of yourself. 🙏 Today's Episode is Brought To you by: Unblocked Your coding agents have access to your codebase, but access doesn't mean good context. Agents can't easily reason across MCPs without guidance — they don't know your architectural decisions, your team patterns, or what that acronym actually means. Unblocked is the context layer your agents are missing. It synthesizes your PRs, docs, Slack messages, and JIRA issues into organizational context so agents make better plans, write higher quality code, use fewer tokens, and require fewer correction loops. Get a free three-week trial at getunblocked.com. 📮 Ask a Question If you enjoyed this episode and would like me to discuss a question that you have on the show, drop it over at: developertea.com. 📮 Join the Discord If you want to be a part of a supportive community of engineers (non-engineers welcome!) working to improve their lives and careers, join us on the Developer Tea Discord community today! 🗞️ Subscribe to The Tea Break We are developing a brand new newsletter called The Tea Break! You can be the first in line to receive it by entering your email directly over at developertea.com. 🧡 Leave a Review If you're enjoying the show and want to support the content head over to iTunes and leave a review!

The Bottleneck Is Moving: Borrowing from traditional manufacturing theory, the coding step used to define your team's total throughput. AI tooling hasn't incrementally improved that bottleneck — it has drastically shrunk it, which means the constraint is now upstream in product decisions, specifications, and prioritization. Engineers who recognize this shift early will redirect their energy accordingly. Sharing Your Opinion Is Not a Free Action: Every time you weigh in on a decision, you're making a transaction. You're asking others to consider your input, and in return, they will update their beliefs about your judgment based on whether you turn out to be right. This means your credibility is a finite resource that appreciates or depreciates over time. Trap #1 — Arguing About Things You Don't Care About: Engineers often feel an intellectual itch to engage when they hear an argument they disagree with, even when the outcome doesn't matter to them. If the only utility of sharing your opinion is your own self-satisfaction, the risk to your social capital almost never justifies the reward. Pick your battles so that when something does matter to you, people actually listen. The Watchful Waiting Approach: If you predict a decision will lead to a bad outcome, sometimes the most effective move is to wait and let the result speak for itself. You get the learning for free without putting your reputation on the line — especially for decisions outside your core responsibilities. Trap #2 — Arguing on the Wrong Axis: When you do engage, make sure your argument is aligned with what the decision-maker actually cares about. A product manager asking engineers to delay optimization work is not going to be moved by arguments about on-call load. An engineering manager introducing a systems design interview won't be swayed by the fact that you personally dislike them. If your reasoning doesn't connect to their criteria, it lands as noise. Naive Realism and the Alignment Fix: We all default to believing our perspective is the balanced, unbiased one. This tendency causes us to assume anyone who disagrees must be missing information. The fix is to start by understanding what the other person is optimizing for. Once you know their criteria, you can either recognize their decision is perfectly reasonable — or reframe your argument in terms they actually care about. The One Takeaway: Understand what the other person wants, what they care about, and why. Decision-making in a collaborative environment is fundamentally negotiation, and the best negotiators optimize for multiple axes rather than treating every disagreement as zero-sum. 🙏 Today's Episode is Brought To you by: Unblocked Your coding agents have access to your codebase, but access doesn't equal context. Agents can't reason across MCPs — they don't know your architectural decisions or why the API is shaped the way it is — so they look in the wrong place and deliver bad outputs. Unblocked is the context layer your agents are missing. It synthesizes your code, PRs, docs, Slack messages, and Jira issues into organizational context that agents actually understand, so they make better plans, write higher quality code, and use fewer tokens. Get a free three-week trial at getunblocked.com/developertea. 📮 Ask a Question If you enjoyed this episode and would like me to discuss a question that you have on the show, drop it over at: developertea.com. 📮 Join the Discord If you want to be a part of a supportive community of engineers (non-engineers welcome!) working to improve their lives and careers, join us on the Developer Tea Discord community today! 🗞️ Subscribe to The Tea Break We are developing a brand new newsletter called The Tea Break! You can be the first in line to receive it by entering your email directly over at developertea.com. 🧡 Leave a Review If you're enjoying the show and want to support the content head over to iTunes and leave a review!

The Power of Physical Checklists: Inspired by aviation, Atul Gawande's The Checklist Manifesto, and Daniel Kahneman's Noise, I've been experimenting with printed, physical checklists for repetitive tasks — from producing this show to running one-on-ones. The rigor of writing precise procedures carries over into clearer communication with both humans and AI agents. Small Interventions, Big Returns: A Brother P-Touch label maker. Reorganizing scattered hobby gear. 3D printing organizational tools with a new Bambu Labs P1S. None of these are revolutionary on their own, but the compounding effect of better organization — essentially building a fast index for your physical life — pays back over and over. Context Shapes Focus: Switching from a home gym to working out at Planet Fitness with my brother-in-law was one of the best focus interventions I've made. The change in environment eliminated the procrastination and context-blending that came from being steps away from my computer. If you're struggling with a habit, sometimes the environment is the variable to change, not your willpower. The Reading List: Good Strategy, Bad Strategy by Richard Rumelt (and its follow-up The Crux), The Art of Action by Stephen Bungay (a great framework for thinking about agentic workflows), How to Know a Person by David Brooks, and my top recommendation: 4,000 Weeks by Oliver Burkeman — a book that will help you stop looking for the productivity hack that fixes everything and start thinking about what actually matters. Learning as a Habit: Right now I'm learning to drive a stick shift on a 1983 Bronco. The point isn't the skill itself — it's staying in the beginner's seat. Intentional practice, setting small goals, refining through repetition. Keeping this habit alive is more important than ever when the industry demands rapid adaptation. How I'm Actually Using AI: Claude Code for one-shotting tools with clear boundaries, local environment improvements, and terminal troubleshooting. OpenClaw for experimental agents like a personalized trip planner and Home Assistant automations via YAML. Claude Co-Work for file system management and screenshot organization. Obsidian as the connective tissue — a markdown knowledge base that gives AI agents personal context to work with. And at work, spec-driven development is showing real promise for shaping agent output quality. A Framework for Thinking About AI's Role: I break AI use cases into categories: automating existing workflows (where most gains are today), operational restructuring (what happens when you free humans from a task), execution of complex technical work (agents on the front lines), iterative consulting on intent and goals, and the emerging frontier of exploratory connections and strategic synthesis. What You Should Actually Do: Be action-oriented — the cat is out of the bag. Invest heavily in planning and specification before sending agents off to work. But more importantly, invest in mindful change: understand your own values, figure out who you want to be when you look back on this moment in 10 years, and let that guide your decisions about adoption, learning, and career direction. 🙏 Today's Episode is Brought To you by: SerpApi If you're building an application that needs real-time search data — whether that's an AI agent, an SEO tool, or a price tracker — SerpApi handles it for you. Make an API call, get back clean JSON. They handle the proxies, CAPTCHAs, parsing, and all the scraping so you don't have to. They support dozens of search engines and platforms, and are trusted by companies like NVIDIA, Adobe, and Shopify. If you're building with AI, they even have an official MCP to make getting up and running a simple task. Get started with a free tier at serpapi.com. 📮 Ask a Question If you enjoyed this episode and would like me to discuss a question that you have on the show, drop it over at: developertea.com. 📮 Join the Discord If you want to be a part of a supportive community of engineers (non-engineers welcome!) working to improve their lives and careers, join us on the Developer Tea Discord community today! 🗞️ Subscribe to The Tea Break We are developing a brand new newsletter called The Tea Break! You can be the first in line to receive it by entering your email directly over at developertea.com. 🧡 Leave a Review If you're enjoying the show and want to support the content head over to iTunes and leave a review!

If you're a software engineer right now, you likely feel like your world is changing overnight. We are writing half or less the amount of code that we wrote even a year ago, which represents a seismic, groundbreaking shift in our industry. However, the rapid introduction of new tools can slide quickly from exciting to purely chaotic, leaving you feeling like you are falling behind. In today's episode, I explore how this changes the nature of our day-to-day work, and why the key to surviving this transition is shifting your mindset from a traditional "Software Engineer" to an "Agent Manager". The Illusion of Velocity vs. Actual Chaos: While the big-picture promise of AI is that the software development pipeline will move exponentially faster, the reality on the ground often feels like unadulterated chaos. Trying to adopt every new tool while spinning up multiple agents to work on parallel tickets introduces a massive new cognitive burden. The Context-Switching Trap: Understand why parallelizing agent workflows fundamentally changes your context-switching overhead. You are no longer just reloading context to build something yourself; you are reloading it to manage, review, and validate a building agent, which rapidly drains your cognitive ability and leads to burnout. The "Agent Manager" Mindset: Treating AI as just a "smart autocomplete" while you try to do the same old job will not work. You need to start viewing your role more like assembly line or process management, focusing on facilitating the system rather than typing every line of syntax. Adopt Old-School Quality Control Tactics: Discover how traditional management methods are becoming essential for individual contributors. Just like a factory manager doesn't inspect every single item off the line, you must develop methods for spot checks, anomaly detection, and standardizing outputs to evaluate the quality and quantity of your agents' work. Shift Your Work Upfront: Recognize that your core effort must move to the specification and planning phases. Your job is increasingly about setting the context, defining the prompt, and establishing strict guardrails before the agent begins its work. Redefining Your Work in Progress (WIP): Proven principles like limiting WIP and focusing on finishing rather than starting are more important than ever to reduce cognitive burden. However, you must adapt these principles to fit a workflow where you are managing processes rather than manually coding. Episode Homework: Take a step back and ask yourself: "What is my true work in progress? Am I actually manually doing these tickets, or am I managing the processes that produce quality ticket work?". 🙏 Today's Episode is Brought To you by: SerpApi No matter what you're building, SerpApi is the web search API for your needs. If you're building an application that needs real-time search data—whether that's an AI agent, an SEO tool, or a price tracker—SerpApi handles it for you. ● Make an API call and get back clean JSON. ● They handle the proxies, CAPTCHAs, parsing, and all the scraping so you don't have to. ● They support dozens of search engines and platforms, and are trusted by companies like NVIDIA, Adobe, and Shopify. ● If you're building with AI, they even have an official MCP to make getting up and running a simple task. Get started with a free tier to build and test your application before you commit. Go to serpapi.com. 📮 Ask a Question If you enjoyed this episode and would like me to discuss a question that you have on the show, drop it over at: developertea.com. 📮 Join the Discord If you want to be a part of a supportive community of engineers (non-engineers welcome!) working to improve their lives and careers, join us on the Developer Tea Discord community today! 🗞️ Subscribe to The Tea Break We are developing a brand new newsletter called The Tea Break! You can be the first in line to receive it by entering your email directly over at developertea.com. 🧡 Leave a Review If you're enjoying the show and want to support the content head over to iTunes and leave a review!