What’s the Meta

Current

1. 64,000 lines down to 9,900

   Cursor Just Showed How Much You’re Wasting On AI · 02:09 Copy link

2. Stop routing models in your head

   Cursor Just Showed How Much You’re Wasting On AI · 02:56 Copy link

3. Could Composer 2.5 do it alone?

   Cursor Just Showed How Much You’re Wasting On AI · 03:43 Copy link

4. Why did Opus barely plan?

   Cursor Just Showed How Much You’re Wasting On AI · 04:53 Copy link

5. Cheap planner, expensive workers

   Cursor Just Showed How Much You’re Wasting On AI · 08:19 Copy link

6. Don’t use frontier intelligence on everything

   Cursor Just Showed How Much You’re Wasting On AI · 09:32 Copy link

7. Cursor’s model router

   Cursor Just Showed How Much You’re Wasting On AI · 10:55 Copy link

8. Robert’s Brain Hands skill

   Cursor Just Showed How Much You’re Wasting On AI · 12:47 Copy link

9. Low reasoning gets almost everything done

   Cursor Just Showed How Much You’re Wasting On AI · 14:12 Copy link

10. GPT 6.1 Sol at a fifth of the price

    Cursor Just Showed How Much You’re Wasting On AI · 16:06 Copy link

11. Overcooking: when more reasoning hurts

    Cursor Just Showed How Much You’re Wasting On AI · 18:24 Copy link

12. Claude finds a new gene editing enzyme

    Opus 5.5 Made Claude Code Usable Again · 32:43 · Watch Copy link

13. Twin check: AI for science

    Opus 5.5 Made Claude Code Usable Again · 37:54 · Watch Copy link

14. Instinct and the personal agent race

    Opus 5.5 Made Claude Code Usable Again · 43:09 · Watch Copy link

15. Agents start spending our money

    Opus 5.5 Made Claude Code Usable Again · 51:59 · Watch Copy link

16. Nvidia's open diarization model

    Opus 5.5 Made Claude Code Usable Again · 56:39 · Watch Copy link

17. Can Jeb replace LLM judges?

    Opus 5.5 Made Claude Code Usable Again · 58:41 · Watch Copy link

18. Twin check: Jeb

    Opus 5.5 Made Claude Code Usable Again · 1:09:04 · Watch Copy link

19. Free output tokens and 200x speed

    Stop Writing Code To Clean Up After Your AI · 32:09 Copy link

20. Jonathan’s early access test

    Stop Writing Code To Clean Up After Your AI · 37:35 Copy link

21. 10,000 API calls: the order of choices matters

    Stop Writing Code To Clean Up After Your AI · 45:11 Copy link

22. Cookbooks: moderation, batching, auto research

    Stop Writing Code To Clean Up After Your AI · 50:16 Copy link

23. Don’t use it for hiring decisions

    Stop Writing Code To Clean Up After Your AI · 1:00:27 Copy link

24. Why it fails at compaction

    Stop Writing Code To Clean Up After Your AI · 1:03:51 Copy link

25. Live demo: the Clarity harness

    Opus 5 Is The First Model That Gets Worse With Use · 1:17:12 Copy link

26. AI memes: Opus 5 over-engineers everything

    Stop Letting Your Best Model Do The Work · 40:44 · Watch Copy link

27. The prompt that fixes Opus 5

    Stop Letting Your Best Model Do The Work · 42:50 · Watch Copy link

28. Grok Bot: multiplayer AI teammates

    Stop Letting Your Best Model Do The Work · 47:58 · Watch Copy link

29. Live demo: Grok Bot

    Stop Letting Your Best Model Do The Work · 50:10 · Watch Copy link

30. Grok Bot as chief of staff for your agents

    Stop Letting Your Best Model Do The Work · 57:23 · Watch Copy link

31. Jeff Dean on how we got here

    Stop Letting Your Best Model Do The Work · 59:50 · Watch Copy link

32. Live demo: the Clarity open source harness

    Your AI Has Amnesia, That’s Why 95% of Deployments Fail · 37:47 · Watch Copy link

33. Building the company brain

    Your AI Has Amnesia, That’s Why 95% of Deployments Fail · 43:37 · Watch Copy link

34. Stacked harness architecture and the LLM judge

    Your AI Has Amnesia, That’s Why 95% of Deployments Fail · 53:00 · Watch Copy link

35. Harness release + waitlist

    Your AI Has Amnesia, That’s Why 95% of Deployments Fail · 1:05:53 · Watch Copy link

36. Live demo: voice + Codex

    OpenAI Lost Control of Its Own Model · 40:30 · Watch Copy link

37. Clarity Aligned: keeping teams aligned

    OpenAI Lost Control of Its Own Model · 42:00 · Watch Copy link

38. Block’s Buzz: the Slack + GitHub killer

    OpenAI Lost Control of Its Own Model · 48:00 · Watch Copy link

39. Spec-driven development with HTML files

    How We Shipped a Production App Using an AI Swarm · 29:58 · Watch Copy link

40. Founder-led sales — Granola to SOW in 15 minutes

    How We Shipped a Production App Using an AI Swarm · 36:23 · Watch Copy link

41. Anatomy of a harness (agent = model + harness)

    How We Shipped a Production App Using an AI Swarm · 52:47 · Watch Copy link

42. What AI builders need to think about right now

    How We Shipped a Production App Using an AI Swarm · 57:37 · Watch Copy link

43. Closed loop agents + slash goal

    How We Shipped a Production App Using an AI Swarm · 1:01:47 · Watch Copy link

44. Elon's 5-step algorithm applied to harness engineering

    How We Shipped a Production App Using an AI Swarm · 1:06:36 · Watch Copy link

45. Verification is the job — the four patterns

    Here's Why Token Maxing Is Bankrupting Companies · 43:05 · Watch Copy link

46. Steering, not typing — patterns 5 through 9

    Here's Why Token Maxing Is Bankrupting Companies · 54:17 · Watch Copy link

47. Vercel Zero: the first agent-native programming language

    The Agent Meta Shift Most Builders Haven't Caught Yet · 00:42:19 · Watch Copy link

48. Codex Desktop: The Visual Feedback Loop

    The $5.5B Proof That APIs Are Dead for Enterprise · 1:05:14 · Watch Copy link

49. OpenAI Codex Pets

    The $5.5B Proof That APIs Are Dead for Enterprise · 1:11:48 · Watch Copy link

50. Claude Code: Managed Agents + Self-Improving Rubrics

    The $5.5B Proof That APIs Are Dead for Enterprise · 1:15:07 · Watch Copy link

51. Claude Code: Dreaming + Persistent Memory

    The $5.5B Proof That APIs Are Dead for Enterprise · 1:25:20 · Watch Copy link