Skip to content
Dalton + MichaelDalton + Michael

How to Build an MVP in the AI Coding Era

Dalton + Michael revisit the classic startup advice around building an MVP and update it for the AI coding era. When building features with Claude Code or Codex is easy, its easy to make something with a ton of features that is bloated and not what people want. A good MVP is just as much about what *not* to build as it is what to build. Remember: don't stop talking to users no matter how tempting it might be. Dalton + Michael is brought to you by @Standard_Cap Sign up for the StandardDB MVP that Dalton references here: https://www.standarddb.com/ Dalton Caldwell on X: https://x.com/daltonc Michael Seibel on X: https://x.com/mwseibel

Michael SeibelhostDalton Caldwellhost
May 18, 202612mWatch on YouTube ↗

CHAPTERS

  1. Why MVP advice must change in the AI coding era

    Michael and Dalton frame the core shift: building features is no longer the primary constraint, so the classic MVP mantra (“build the smallest valuable thing because building is expensive”) needs updating. The new environment makes it easy to build too much too soon, which can actively harm learning and adoption.

  2. AI-driven feature creep arrives in days, not months

    Dalton shares a recent launch experience where AI-enabled speed led to an overbuilt MVP. The surprising new work wasn’t building—it was deleting and simplifying to regain clarity.

  3. Founders must become editors: cut scope to create value

    They argue the key skill has shifted from implementation to editing—choosing what not to build. Broad surface-area products created by “vibe coding” often produce confusion rather than value.

  4. The biggest trap: building for others without user reality checks

    Michael explains how infinite build capacity is especially dangerous when the founder isn’t the user. Without tight feedback loops, it’s easy to rationalize imagined needs and postpone real user conversations.

  5. Don’t implement the user’s feature list—dig for the real problem

    They warn that AI makes it trivially easy to implement every requested feature, but that’s a path to a “tar pit.” Instead, founders must interpret requests and uncover underlying pain points and priorities.

  6. User-pull, not tool-push: expand only when demand forces it

    Dalton emphasizes adding features should be a response to strong user pull, not a reflection of what AI can generate. The goal is to avoid complexity until it’s clearly justified by real usage and retention.

  7. Talking to users vs spamming them: AI amplifies the wrong behavior

    They reaffirm the importance of talking to users, but note that AI also makes spam outreach and lead-buying easier than ever. High-volume outreach can create misleading signals and low-quality learning.

  8. Fewer, deeper conversations create real insight and word-of-mouth

    Michael argues startups aren’t just “ask users, build what they say.” Customers often don’t fully know what will make them successful, so founders need intimate understanding to actually move the needle—leading to authentic referrals.

  9. Case study: 12 Zoom calls and the counterintuitive power of less

    Dalton recounts conducting a dozen high-quality calls with potential users. The product was initially too complex to understand; removing most functionality made it immediately actionable and drove instant signups.

  10. GarageBand metaphor: democratized creation doesn’t increase demand

    Dalton compares AI coding to music production tools: it’s easier than ever to create “radio-quality” output, but demand and attention don’t scale. Most output becomes “slop,” while winners remain rare and highly skilled.

  11. Where AI may help most: empowered insiders vs breakout startups

    They note AI tools can be exceptionally powerful for people with clear goals inside companies—boosting productivity without needing to find external demand. Building internal tools/content has a larger addressable base than producing “blockbusters.”

  12. Build in public, but don’t publish AI-generated slop

    Dalton updates the “build in public” advice: AI makes it easy to flood LinkedIn/X with generic content, which blends into the noise. Distinctive, expertise-driven writing builds real trust and audience.

  13. Avoid fake momentum: positive-looking loops can hide negative churn

    Michael describes how AI-enabled spam can create the appearance of traction—more outreach, more signups—while the underlying product fails and users churn. This creates an opportunity for disciplined teams to differentiate.

  14. The new MVP north star: clarity, focus, and ‘two features that work’

    They close by arguing that in an AI world, the winning strategy is selective simplicity. A small number of features that clearly improve a user’s life beats a sprawling product that initially impresses but ultimately confuses.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

Add to Chrome