Lennox Saint
Open menu
the show

less scrolling.
more “oh shit, i could do that”

a weekly show for people building with AI. the catch-up, then one thing i actually tried. the prompt is in the field notes.

free. unsubscribe any time, one click. privacy.

watch or listen anywhere

LAB · trailer · 26s

every episode · field notes included

pick one. steal the prompt.

29 episodes so far. a new one every week.

  1. LAB 0028new episode · 24 oct 2026Claude is quietly testing new limits - i checked 5 of my accountswatch + copy the promptthe newest one
  2. LAB 0027new episode · 24 oct 2026PHP got hacked wide open this week - one Claude prompt tells you if you're hitwatch + copy the prompt
  3. LAB 0026new episode · 22 oct 2026Claude made 2 of these Shorts from one prompt. can you tell which?watch + copy the prompt
  4. LAB 0025new episode · 22 oct 2026Claude Slides vs ChatGPT Slides - the loser costs me my ChatGPT planwatch + copy the prompt
  5. LAB 0024new episode · 20 oct 2026devs are paying double for the old GPT - i tested if they're rightwatch + copy the prompt
  6. LAB 0023new episode · 19 oct 2026Meta just gave away its best model for free - i tested the catchwatch + copy the prompt
  7. LAB 0022new episode · 16 oct 2026Anthropic is so far ahead it's not even funny - Claude now runs a team of agentswatch + copy the prompt
  8. LAB 0021new episode · 15 oct 2026Meta and Google both say they caught Opus - so i tested themwatch + copy the prompt
  9. LAB 0020new episode · 14 oct 2026X just merged Grok and Cursor into one plan - but there's a catchwatch + copy the prompt
  10. LAB 0019new episode · 12 oct 2026Codex now writes your next prompt - biggest AI change this yearwatch + copy the prompt
  11. LAB 0018new episode · 10 oct 2026GPT-6.1 Sol Ultrafast: up to 8x faster for $500watch + copy the prompt
  12. LAB 0017new episode · 9 oct 2026Elon's Grok Bot will use Claudewatch + copy the prompt
  13. LAB 0016new episode · 8 oct 2026OpenAI's rival called their 28 days "nothingburgers"watch + copy the prompt
  14. LAB 0015new episode · 8 oct 2026Claude gives you 5x more than ChatGPT (now it's measured)watch + copy the prompt
  15. LAB 0014new episode · 7 oct 2026Codex usage resets are a casino - plus GPT-6.1 Sol on BuilderBenchwatch + copy the prompt
  16. LAB 0013new episode · 5 oct 2026Fable 5.5 is leaking, and Claude is already pulling awaywatch + copy the prompt
  17. LAB 0012new episode · 3 oct 2026ChatGPT wants your whole day - so i let it audit minewatch + copy the prompt
  18. LAB 0011new episode · 1 oct 2026OpenAI's $500 plan feels like the old $200 plan - but dots are freewatch + copy the prompt
  19. LAB 0010new episode · 30 sep 2026i gave Sonnet 5.5 4 hours of my real work. it scored 48.98watch + copy the prompt
  20. LAB 0009run costs · 28 sep 2026score a new model on your own benchmarkwatch + copy the prompt
  21. LAB 0008effort levels · 28 sep 2026start on high effort, not maxwatch + copy the prompt
  22. LAB 0007blind review · 26 sep 2026judge AI work blindwatch + copy the prompttested
  23. LAB 0006benchmark · 25 sep 2026build a benchmark for your real workwatch + copy the prompttested
  24. LAB 0005brand · 24 sep 2026give it your brand firstwatch + copy the prompt
  25. LAB 0004build · 23 sep 2026feedback becomes softwarewatch + copy the prompt
  26. LAB 0003inbox · 21 sep 2026teach AI your inbox ruleswatch + copy the prompt
  27. LAB 0002voice to tasks · 20 sep 2026turn a voice dump into separate work laneswatch + copy the prompt
  28. LAB 0001feedback loops · 18 sep 2026stop giving your AI the same feedbackwatch + copy the prompt
  29. LAB 0000focus · 17 sep 2026get your agent back to the actual jobwatch + copy the prompt

inside every episode

two parts. one thing to try.

part one

less scrolling

the AI news worth your time, with the source and the limits.

part two

more oh shit, i could do that

one real job i handed to AI. the result, the rough bits, the fix.

what it looks like on the show

every idea gets drawn, not just said.

real graphics from recent episodes. they play as you scroll.

  • LAB 0007
  • LAB 0008
  • LAB 0009
  • LAB 0009

then

the field notes

the steps, the prompt to copy, and the checks that caught my mistakes. free, every episode.

open the latest
BuilderBench by the LABBuilderBench by the LABi gave AI my real job.same jobs. same clock. scores locked, limits published. Gemini 4 Argon 45.38 (pack 3.0.0), Muse Wonder Full 39.91 (pack 3.0.0), GPT-6 Luna 44.58 (pack 3.0.0), Haiku 5.5 39.18 (pack 3.0.0), GPT-6.1 Sol 41.75 (pack 3.0.0), Sonnet 5.5 48.98 (pack 3.0.0), GPT-6 Astra 53.75 (pack 2.0.0), Opus 5.5 68.26 (pack 1.0.0), GPT-6 Sol 45.55 (pack 1.0.0).see the scorecard
Lennox on camera, with the LAB lower third: Lennox Saint, AI builder

your host quit law the day he graduated.

now he tests AI on real work, in public, and shows the receipts.

the short version of the story

p.s.

get the field notes for every episode.

  • the exact prompt i typed
  • what came back
  • where it fell over

free. unsubscribe any time, one click. privacy.

  • 32,718 on Threads
  • 3,675 readers
  • 1.84K on YouTube

live count, 24 oct 2026 / 25 oct 2026.

running a team that loses hours to repeated work? see how i help.