Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Developer Tea

Developer Tea artwork

Why Can't You Go Faster With AI? Focus on the Friction to Find Out

Jun 24, 202619 minAudio

If you are a manager, a lead engineer, or anyone growing into more responsibility, this throwback episode is built for you. We keep hearing the same question, now louder than ever: "Why can't this go faster?" AI and agentic coding have made the literal coding step dramatically cheaper, so product leaders reasonably expect the whole pipeline to speed up. But it hasn't - and in today's short, focused episode I explore why. The answer isn't new at all. It's the theory of constraints, and it has everything to do with friction you may not be looking at. Speed Isn't the Story - Friction Is: When a fast component gets introduced into the pipeline, the instinct is to celebrate the velocity. But pay attention to what comes after. The real question is what keeps work from naturally flowing faster, and that lives in the friction, not the energy you're pouring in upfront. The Universal Bottleneck: I rarely claim universal truths on this show, but here's one: anything that looks like a pipeline will have a bottleneck. If you're not paying attention to it, it doesn't matter how fast every other step gets. Faster coding just exposes where the constraint really sits. The Two Places Friction Shows Up: For teams fully adopting agentic coding, the bottlenecks cluster in two spots - requirements gathering at the front, and verification, validation, and testing at the back. Rushed requirements upstream create even more painful rework downstream. Why Agents Punish Vague Specs: Human engineers fill in gaps by being close to the work. Agents fill in gaps too, but sometimes incorrectly. If your requirements aren't detailed, the agent guesses, and you pay for it in review. Spend more time in the planning phase, not less. The Foundation You Build On: Agents glob extra code onto a weak structure - unnecessary models, redundant endpoints, patterns that don't fit. A code base organized with clear conventions, good documentation in your CLAUDE.md or AGENTS.md, and dependable patterns lets the agent discover and extend rather than guess and hope. Specification and Validation Are Bookends: Good requirements translate directly into good tests. Acceptance criteria on one end, changes in the middle, validation on the other end - directly connected. Poor specification sitting on a poor structure guarantees poor execution and poor validation. Reframing Your Objections: Think scope creep is the problem? That's a requirements issue. Think you lack the talent? That's a foundation issue - because the engineer's job now is to cultivate the foundation so generated code enriches it instead of toppling it. This is not a new problem. We asked it of the internet, of web frameworks, of CSS. Now it's time to apply the same principles to agentic velocity: look at your requirements, your foundation, and your validation. Somewhere in those three is your bottleneck. I guarantee it. ๐Ÿ“ฎ 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. ๐Ÿ“ฎ 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! ๐Ÿ—ž๏ธ 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. ๐Ÿงก Leave a Review If you're enjoying the show and want to support the content head over to iTunes and leave a review!

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. What Kind of Engineers Are Getting Hired Right Now?

    Nov 20, 202615 minAudio

    What kind of engineers are getting hired today? Some of what hiring managers look for hasn't changed at all, but the reasons it matters have shifted, and the importance of those traits has grown. In this episode I talk...

    Credit the people on this episode

  2. Finding A Way Through Forced Change of AI Transformations

    Nov 14, 202618 minAudio

    You likely chose most of the changes in your career yourself, and many engineers now feel AI-driven change is being forced on them. In this episode, I talk about why that shift feels so personal, and how three ideas can...

    Credit the people on this episode

  3. Three Default Thinking Modes That Can Screw Up Your Planning and Vision

    Nov 5, 202617 minAudio

    Your brain is constantly trying to simplify the world. It takes a snapshot of what it sees, builds a picture that makes sense, and then works from that picture. Most of the time this works well enough. But when we plan...

    Credit the people on this episode

  4. Signs You May Have a Regressive Mindset

    Oct 24, 202617 minAudio

    Keeping up has always been part of the job for software engineers, and with AI it's harder than ever. In this episode, I talk about two signals that you, your coworkers, or the people you manage might have a regressive...

    Credit the people on this episode

  5. How History Can Inform a Dynamic Understanding of Technological Revolutions Mirroring this Moment In Software Engineering

    Sep 30, 202617 minAudio

    You're tired of hearing that you need to pick up AI, so today I'm not going to add to that pile. Instead, we'll look back at two moments in history that look a lot like this one: the arrival of Fortran in the 1950s, and...

    Credit the people on this episode