Videos xUnRQ9vLXxo

Everything we knew about software has changed — Theo Browne, @t3dotgg

index_state ready data_status ok

AI Engineer· published 2026-07-08· 0:16:01· en-US· indexed 2026-08-10 19:46

Open on YouTube

40 shot(s).

keyframes kept every frame deduplicated

What was stored

cues: 224; whisperx 224
chunks: 29; from 224 cues
keyframes: 18; kept of 40 captured
frames with text: 18; 345 lines read
chapters: 9; from the source metadata
keyframe bytes: 3.7 MB; word timings on 224 cues

Provenance

Each pipeline stage, its state and the model that produced it
stage	state	model	started	took
`fetch`	done	—	2026-08-10 01:49	1m 41s
`stt`	done	—	2026-08-10 01:51	18s
`chunk`	done	—	2026-08-10 01:51	0s
`text_embed`	done	—	2026-08-10 19:46	1s
`keyframe`	done	—	2026-08-10 01:51	1m 37s
`ocr`	done	—	2026-08-10 01:52	5s
`frame_embed`	done	—	2026-08-10 19:46	3s

Frames, and what the machine read

0:02 #0 done2 line(s)

shot 0·sharpness 455.1
1. AlEngineer0.96
2. World's Fair0.97
0:03 #1 done2 line(s)

shot 1·sharpness 656.7
1. AlEngineer0.96
2. World's Fair0.99
0:10 #2 done24 line(s)

shot 2·sharpness 2734.3
1. LAB & PLATINUM SPONSORS0.99
2. Amazon AGI Lab0.98
3. ANTHROP\C1.00
4. Google DeepMind1.00
5. MINIMAX0.93
6. OpenAI0.93
7. Akamai1.00
8. arize1.00
9. aws1.00
10. Braintrust bright data0.99
11. B1.00
12. Browserbase1.00
13. docker1.00
14. :neo4j0.92
15. ORACLE1.00
16. PayPal1.00
17. qodo1.00
18. reducto1.00
19. Sonar1.00
20. Makers of0.99
21. togetherai1.00
22. Unblocked1.00
23. WorkOS1.00
24. SonarQube1.00
0:15 #3 done4 line(s)

shot 3·sharpness 674.3
1. THEO BROWNE0.99
2. FOUNDER1.00
3. t3.gg1.00
4. AIE1.00
0:20 #4 done85 line(s)

shot 4·sharpness 2967.6
1. World's Fair0.98
2. OINNGEST0.93
3. World'sFair1.00
4. AlEngineer0.97
5. BAND1.00
6. World'sFair1.00
7. Cleric1.00
8. World'0.99
9. AlEngin0.98
10. Snorkel0.92
11. World's Fair0.98
12. Z.AI0.98
13. World's Fair0.98
14. qodo0.99
15. World's Fair0.98
16. AlEngineer0.98
17. Payl0.96
18. World's Fair0.95
19. AlEngineer0.99
20. neo4j0.99
21. World'sFair1.00
22. AlEngineer0.99
23. Google DeepMind1.00
24. World's Fair0.99
25. AlEngineer0.98
26. arize1.00
27. World's0.96
28. AlEngin0.99
29. WorkOS1.00
30. World's Fair0.95
31. AlEngineer1.00
32. Amazon AGI Lal0.99
33. World's Fair0.98
34. AlEngineer0.98
35. Microsoft1.00
36. World's Fair0.98
37. ORA1.00
38. AlEngineer0.98
39. AlEngin0.98
40. World's Fair0.99
41. Braintrust1.00
42. World'sFair1.00
43. enAl0.99
44. World'sFair1.00
45. MINIMAX1.00
46. World's0.96
47. aws1.00
48. World's Fair0.99
49. AlEngineer1.00
50. togethera.1.00
51. 1d's Fair0.94
52. Akamai1.00
53. World's Fair0.97
54. Unbl0.99
55. World's Fair0.99
56. AlEngineer0.99
57. World'sFair1.00
58. AlEngineer0.99
59. ATADOG1.00
60. World'sFair1.00
61. AlEngineer0.99
62. Resolve.ai1.00
63. World'0.95
64. AlEngin0.99
65. CLOUDFLARE1.00
66. Ingineer0.99
67. AlEngineer0.99
68. LanceDB1.00
69. World's Fair0.98
70. builder.io1.00
71. d'sFair1.00
72. Ravenna1.00
73. World's Fair0.96
74. Copilot1.00
75. World's Fair0.98
76. AlEngineer1.00
77. TOPK1.00
78. World'sFair1.00
79. AlEngineer0.98
80. gnee1.00
81. World'sFair1.00
82. eNCORD0.97
83. World'0.94
84. AlEngin0.93
85. Al0.89
0:50 #5 done13 line(s)

shot 5·sharpness 717.9
1. AIEngineer0.96
2. I Lab0.96
3. World's Fair1.00
4. Microsoft1.00
5. AlEngineer0.97
6. Oper1.00
7. air1.00
8. Vorld's Fair0.92
9. AIE1.00
10. erai1.00
11. Worlus0.98
12. Akamai1.00
13. AlEngineer0.97
1:13 #6 done14 line(s)

shot 6·sharpness 1559.8
1. AlEngineer0.98
2. World's Fair0.97
3. PRESENTED BY1.00
4. Microsoft1.00
5. Sonnet3.51.00
6. - AlEngineer0.96
7. 引l Lab0.80
8. prld'sFair0.99
9. -air0.84
10. 1Al0.73
11. Keynote1.00
12. era.0.99
13. 'sFair0.92
14. Theo Browne / Founder t3.gg0.98
1:40 #7 skipped

shot 7·duplicate of #6
2:13 #8 skipped

shot 8·duplicate of #6
2:38 #9 done19 line(s)

shot 9·sharpness 2177.2
1. AlEngineer0.98
2. World's Fair0.97
3. Sonnet 3.51.00
4. Tool calls1.00
5. PRESENTED BY1.00
6. Microsoft1.00
7. Opus 4.50.98
8. Long running work1.00
9. Mythos 50.99
10. Orchestration1.00
11. Micros1.00
12. Wi0.73
13. AlEnginer0.98
14. World'1.00
15. Keynote1.00
16. Aka0.99
17. W1.00
18. Theo Browne / Founder t3.gg0.98
19. AlEng0.97
2:55 #10 skipped

shot 10·duplicate of #9
3:36 #11 done12 line(s)

shot 11·sharpness 1746.5
1. AlEngineer0.99
2. World's Fair0.97
3. PRESENTED BY1.00
4. We need to get over ourselves.1.00
5. Microsoft1.00
6. AlEngineer1.00
7. World'sFa1.00
8. Op1.00
9. Woi0.82
10. Keynote1.00
11. Wor0.89
12. Theo Browne / Founder t3.gg0.99
4:01 #12 skipped

shot 12·duplicate of #11
4:36 #13 done21 line(s)

shot 13·sharpness 1947.0
1. AlEngineer1.00
2. World's Fair0.99
3. 9:42 AM1.00
4. 315° NW0.93
5. PRESENTED BY1.00
6. Microsoft1.00
7. N0.80
8. 3S0.76
9. 3719'55"N, 122°1°46°W0.88
10. d'sFair0.99
11. ngineer1.00
12. Micros1.00
13. enAl0.95
14. rld'sF0.99
15. AlEngineer -0.94
16. ngineer1.00
17. Keynote1.00
18. d'sFal0.90
19. Akama1.00
20. Theo Browne / Founder t3.gg0.98
21. AlEngineer -0.94
5:05 #14 done31 line(s)

shot 14·sharpness 2061.3
1. AlEngineer0.99
2. World's Fair0.97
3. * AT&T M-Coll 6:45 AM0.91
4. 2661.00
5. PRESENTED BY0.98
6. 3001.00
7. 1501.00
8. Microsoft1.00
9. 3301.00
10. 1200.92
11. E1.00
12. 901.00
13. 601.00
14. 301.00
15. 228°0.98
16. SW0.95
17. Boulder1.00
18. Colorado1.00
19. 40°3'27°N 105°13'8* W0.91
20. — AlEngineer0.92
21. orld's Fair0.97
22. Mic1.00
23. -AlEngi0.94
24. Open/0.83
25. 'orld0.99
26. - AlEnginee0.94
27. Keynote1.00
28. orld'sta0.95
29. Aka0.93
30. Theo Browne / Foundert3.gg0.98
31. AlEngi0.97
5:29 #15 skipped

shot 15·duplicate of #14
5:47 #16 skipped

shot 16·duplicate of #14
6:18 #17 skipped

shot 17·duplicate of #9
6:42 #18 skipped

shot 18·duplicate of #6
7:03 #19 skipped

shot 19·duplicate of #6
7:50 #20 skipped

shot 20·duplicate of #11
8:11 #21 skipped

shot 21·duplicate of #11
8:25 #22 skipped

shot 22·duplicate of #11
8:54 #23 skipped

shot 23·duplicate of #11

Transcript

224 cues· 3,198 words· 16,610 chars

chunk 0 · 0:17–1:00 · 150 words · 778 chars
0:17 Hello, hello, fantastic to see you guys here.
0:20 I still can't believe they're letting me take a stage at something like this, a YouTuber apparently, but can't wait to share a bit about how I've been thinking because if I'm being real, kinda going through some AI psychosis.
0:33 Who here would classify how they feel right now as some form of AI psychosis?
0:37 I wanna see some hands.
0:40 Those who don't have their hands up yet, don't worry, we'll get you there by the end of this talk if I do everything right.
0:45 In order to talk about this, I wanna start with a bit of a personal journey of my own, and I'm gonna go through this the way anybody does in modern timelines, with the models.
chunk 1 · 0:54–1:39 · 142 words · 769 chars
0:54 Who here used Sonnet 3.5 when it was the creme de la creme, the cream of the crop model available to us?
1:01 It was unbelievably better than what we had used before, right?
1:05 Like having used all of these different models and trying them in tools.
1:08 Sonnet 3.5 was a big moment for me because it felt like these models could suddenly complete much more end-to-end tasks, like actually get real work done that takes multiple steps.
1:18 And then we got Opus 4.5.
1:19 I'll go differently here.
1:23 Who didn't feel a big jump when they switched over to Opus 4.5?
chunk 2 · 1:28–2:09 · 142 words · 745 chars
1:28 That's a relief, there were not too many hands because Ovis 4.5 is probably when my psychosis started in November and December of last year.
1:34 Having a model that couldn't just write the code and call tools but could go way further.
1:39 A model that could test the work and actually get it into a good state and complete tasks that take hours instead of minutes.
1:46 It was unbelievable.
1:48 And then we got Mythos.
1:49 Who here has had a chance to play with Mythos and Fable so far?
1:54 We agree it's a pretty damn good model, right?
1:57 But why?
chunk 3 · 1:58–2:38 · 157 words · 806 chars
1:58 It's not just better at coding.
1:59 If you handed a prompt that you would have handed to these other models before, it's not gonna feel that different.
2:05 I think of these almost as eras now, where Sonnet 3.5 is the tool call era.
2:09 Not that it was the first model that could do tool calls, rather it was the first one that did them consistently and reliably enough in context of a code base where you could use this for day-to-day coding work.
2:21 Then we got Opus 4.5, which was able to do much longer running tasks without losing track of what it's working on.
2:26 It's no longer, okay, build step one, and then it does it.
chunk 4 · 2:28–3:12 · 172 words · 904 chars
2:28 Then you say, okay, can you build this next part, and then the next part.
2:31 You can just tell it what you want, and it could figure it out a lot of the time.
2:35 Mythos is another jump to orchestration.
2:39 It feels to me like it's the first model that doesn't just understand your code base, but it understands itself.
2:44 And it knows how to spawn additional models and break up work in a way where it can be completed more reliably and then verified afterwards.
2:51 And if you tell the model to do that, it will just do it.
2:53 You don't need some custom tooling, some custom systems, some fancy software factory.
2:58 You just need to prompt it to go a little further.
chunk 5 · 3:00–3:42 · 159 words · 836 chars
3:00 I think you'll be surprised how far it can go.
3:03 What I'm trying to say here is we need to go bigger.
3:06 You're not going to see the benefits going forward if you're not pushing the model further, and you're not pushing yourself further with what you're building.
3:12 Most of the Jira tickets I closed at my previous job could be trivially solved with a model like Opus 4.5.
3:18 My previous work would not benefit from a model like Mythos.
3:22 If the models are gonna keep getting better, and at this point I'm confident in saying they are.
3:26 I was wrong when I claimed that we were hitting a wall before.
3:29 The models are getting better faster than we are, so we can't necessarily get better, so instead we have to go bigger.
chunk 6 · 3:35–4:16 · 149 words · 737 chars
3:35 In order to do that, we have to get over ourselves.
3:37 This was really hard for me as someone who spent a long time writing software.
3:42 Who here has written code for more than 10 years?
3:45 I wanna see hands.
3:46 It's the majority of the people here.
3:49 I don't even want to think about how long I've been writing code for.
3:51 But I have been building up all of these strong opinions since I started.

Chapters

0:00 Introduction and the "AI psychosis" experience
0:50 Evolution of AI models: Sonnet 3.5, Opus 4.5, and Mythos
3:04 The imperative to "go bigger" and push model capabilities
3:35 Overcoming legacy constraints and developer habits
6:08 Moving past our "skeuomorphic" phase in software development
9:30 Personal project evolution: side projects, startups, and the "Markdown tier"
12:22 Identifying the gap: What is "too big" anymore?
13:06 Redefining the strategy: Building for a wider spectrum instead of just depth
14:50 Scaling and architecting products to allow for user-driven extensibility