$30 off During Our Annual Pro Sale. View Details »

If we only had a brain

If we only had a brain

AI is having a major impact in how well we work together and how we develop software. But is it a positive impact?

In this talk I review how there's been little improvement in outcomes of software development project over time, and how the way most are using generative AI will only continue that trend and have no positive impact whatsoever.

The talk ends with some explicit advice on how to use LLMs to enable the human and enhance communication.

Avatar for Wouter Lagerweij

Wouter Lagerweij

November 24, 2025
Tweet

More Decks by Wouter Lagerweij

Other Decks in Technology

Transcript

  1. [email protected] “We build systems like the Wright brothers built airplanes

    — build the whole thing, push it off the cliff, let it crash, and start over again.”
  2. [email protected] “Production of large software has become a scare item

    for management. By reputation it is often an unprofitable morass, costly and unending. This reputation is perhaps deserved.”
  3. [email protected] “The most deadly thing in software is the concept,

    which almost universally seems to be followed, that you are going to specify what you are going to do, and then do it. And that is where most of our troubles come from.”
  4. [email protected] Success comes from • Strong executive sponsorship • Clear,

    well-defined requirements • Agile and adaptive project management practices • Robust communication channels • Effective risk and change management processes
  5. [email protected] “Users are interested in systems requirements and buy systems

    in that way. But that implies that they are able to say what they want. Most of the users aren’t able to.”
  6. [email protected] https://news.mit.edu/2025/mit-researchers-propose-new-model-for-legible-modular-software-1106 Old Is the New New • Kevlin Henney

    • GOTO 2018 “Their new approach breaks systems into “concepts,” separate pieces of a system, each designed to do one job well, and “synchronizations,” explicit rules that describe exactly how those pieces fit together.”
  7. [email protected] “The distinction between design and production is essentially a

    practical one, imposed by the need for a division of the labor. In fact, there is no essential difference between design and production”
  8. [email protected] 20 days 32 days 21 days 3 days 7

    days 1 day 2 days 2 days 3 days 0,5 day =
  9. [email protected] 20 days 32 days 21 days 3 days 7

    days 1 day 2 days 2 days 3 days 0,5 day 81 days 8,5 days = 10%
  10. [email protected] 20 days 32 days 21 days 3 days 7

    days 1 day 2 days 2 days 3 days 0,5 day 81 days 8,5 days = 10%
  11. [email protected] Despite $30–40 billion in enterprise investment into GenAI, this

    report uncovers a surprising result in that 95% of organizations are getting zero return Tools like ChatGPT and Copilot are widely adopted. Over 80 percent of organizations have explored or piloted them, and nearly 40 percent report deployment. But these tools primarily enhance individual productivity, not P&L performance. Meanwhile, enterprise-grade systems, custom or vendor-sold, are being quietly rejected. Sixty percent of organizations evaluated such tools, but only 20 percent reached pilot stage and just 5 percent reached production. Most fail due to brittle workflows, lack of contextual learning, and misalignment with day-to-day operations
  12. [email protected] Despite a surge in generative AI use across workplaces,

    most companies are seeing little measurable ROI. One possible reason is because AI tools are being used to produce “workslop”—content that appears polished but lacks real substance, offloading cognitive labor onto coworkers. https://hbr.org/2025/09/ai-generated-workslop-is-destroying-productivity
  13. [email protected] Example: The Dev Loop • Optimize tooling for humans

    • eXtreme Vibe Programming • Glamorous Toolkit / Moldable Development Be in the loop
  14. [email protected] Example: Learning over generation • Gather options and ideas

    • Validate • Communicate Don’t send your colleagues generated text. More is not better! Be in the loop
  15. [email protected] • Make use of shorter cycles, include the business

    • Maybe that team can be smaller? Example: Come together Optimize for fast feedback
  16. [email protected] Example: Standardization • An old idea is not a

    bad idea… • Publish into to model Optimize for communication
  17. [email protected] Example: Spec together • Fast generation of examples and

    scenarios • Fast feedback, common understanding • Straight into spec-driven programming Optimize for communication
  18. [email protected] Example: Empathy • Are you offloading your understanding on

    others? • Is the message clearer, or just longer? Consider the system