Speaker Notes

Speaker Notes

190 talks · 50 channels 211h of video · 8h to read

Notes on long developer talks: summaries, highlights and answers, each cited to the second

Talk

AI-Friendly Code: The Missing Link for Speed with Quality

Coding agents dramatically increase developer speed but degrade code health and skyrocket defects unless quality rules and MCP safeguards are enforced.

Adam Tornhill · GOTO Copenhagen 2026 · GOTO Conferences · · 2 min read · 45 min talk

Read the notes · 2 min or watch 15 min of highlights

In this talk

  • 4:34AI coding tools increase defects and complexity
  • 11:21AI agents have much higher defect rates on unhealthy code
  • 19:11Safeguarding AI code using an MCP code health server

It answers

  • 5:31Why do initial AI velocity gains disappear after a couple of months?
  • 9:30How does code health impact task completion speed and defect risk?
  • 12:02Are coding agents more sensitive to code quality than human developers?

Quote of the week

An agent can't be fired or sued or embarrassed, and it can't vouch for any software.

Kent C. Doddsin How I'd Get Into Software in 202610:49

Start here

Three talks that explain the field

Coding agents

All 70 Coding agents talks →

Models & infra

All 50 Models & infra talks →
Local models

Agentic coding workflows with small models

Small language models complement frontier models in developer workflows, offering cost-efficient local execution when properly configured with quantization, expanded context, and task orchestration.

Android Developers · · 1 min read · 32 min talk

Teams & work

All 58 Teams & work talks →
Developer careers

How I'd Get Into Software in 2026

To thrive in software engineering in 2026, developers must shift from writing raw code to communicating intent, leveraging the adjacent possible, and offering human accountability and deep domain expertise.

Kent C. Dodds · · 1 min read · 15 min talk

Every talk, read in minutes

Each talk gets notes you can read in a couple of minutes, and every line points back to the moment it comes from.

  • SummaryOne paragraph per part of the talk, each with its time. 0:12 AI coding speeds up work but risks quality
  • HighlightsThe moments that matter, played back to back: 15 min of a 45 min talk.
  • AnswersThe questions it answers, in the speaker's words. “Why do initial AI velocity gains disappear after a couple of months?”

The whole library, three ways

190 talks from 50 channels

Reading takes a fraction of watching, and every note links to the second it came from.

From 50 channels

AI Native Dev37 talks AI Engineer32 talks Dwarkesh Patel14 talks The Pragmatic Engineer13 talks Latent Space6 talks Andrej Karpathy5 talks Machine Learning Street Talk5 talks Y Combinator5 talks GOTO Conferences4 talks Lex Fridman4 talks Sequoia Capital4 talks a16z3 talks Fireship3 talks Lenny's Podcast3 talks 3Blue1Brown2 talks AI Engineer and Matt Pocock2 talks and 34 more →