Videos W76woOYHlvY

Software Engineering Is Becoming Plan and Review — Louis Knight-Webb, Vibe Kanban

index_state ready data_status ok

AI Engineer· published 2026-05-02· 0:20:23· en-US· indexed 2026-08-10 19:54

Open on YouTube

72 shot(s).

keyframes kept every frame deduplicated

What was stored

cues: 232; whisperx 232
chunks: 34; from 232 cues
keyframes: 52; kept of 72 captured
frames with text: 52; 2,056 lines read
chapters: 10; from the source metadata
keyframe bytes: 7.7 MB; word timings on 232 cues

Provenance

Each pipeline stage, its state and the model that produced it
stage	state	model	started	took
`fetch`	done	—	2026-08-10 18:07	1m 54s
`stt`	done	—	2026-08-10 18:09	21s
`chunk`	done	—	2026-08-10 18:10	0s
`text_embed`	done	—	2026-08-10 19:54	1s
`keyframe`	done	—	2026-08-10 18:10	1m 52s
`ocr`	done	—	2026-08-10 18:12	30s
`frame_embed`	done	—	2026-08-10 19:54	9s

Frames, and what the machine read

0:03 #0 done2 line(s)

shot 0·sharpness 658.4
1. AlEngineer0.98
2. EUROPE1.00
0:08 #1 done2 line(s)

shot 1·sharpness 827.9
1. PRESENTINGSPONSOR1.00
2. Google DeepMind1.00
0:13 #2 done3 line(s)

shot 2·sharpness 907.7
1. PLATINUM SPONSORS0.98
2. # Braintrust0.96
3. WorkOS OpenAI0.95
0:21 #3 done7 line(s)

shot 3·sharpness 603.6
1. AlEngineer0.98
2. EUROPE1.00
3. What are we even0.98
4. to do all day?1.00
5. Software Engineer Edition1.00
6. AlEngineer0.99
7. EUROPE1.00
0:37 #4 done11 line(s)

shot 4·sharpness 1827.7
1. *★*0.53
2. AIE1.00
3. ★1.00
4. What are we even0.97
5. going1.00
6. to do all day?1.00
7. Software Engineer Edition1.00
8. Google DeepMind1.00
9. AlEngineer0.99
10. EUROPE1.00
11. 20261.00
0:50 #5 done10 line(s)

shot 5·sharpness 2323.2
1. AIE1.00
2. Louis Knight-Webb0.97
3. @tokengobbler1.00
4. ★1.00
5. vibekanban.com1.00
6. london.aitinkerers.org1.00
7. Engineering the future of Al0.99
8. AlEngineer0.99
9. EUROPE1.00
10. 20260.99
1:08 #6 done91 line(s)

shot 6·sharpness 2080.1
1. Leaderboards1.00
2. There's an all-nw, challenging SWE-bench Multimodal, containing software issues described with images. Leam more here.0.94
3. Bash Only0.99
4. Verified1.00
5. Lite1.00
6. Full1.00
7. Multimodal1.00
8. Verified is a human-filtered subset of 500 instances (details)0.99
9. Fiters:0.99
10. No Filters0.97
11. All Tags0.96
12. Model1.00
13. % Resolved0.99
14. Org1.00
15. Date1.00
16. Site1.00
17. EPAM AI/Run Developer Agent v20250719 + Claude 4 Sonnet0.98
18. 76.801.00
19. 2025-08-041.00
20. e0.65
21. ★1.00
22. ACoder1.00
23. 76.401.00
24. 2025-08-191.00
25. e0.86
26. AIE1.00
27. ★1.00
28. TRAE1.00
29. 75.200.99
30. 2025-06-121.00
31. e0.96
32. ★1.00
33. Harness AI1.00
34. 74.801.00
35. 2025-07-311.00
36. e0.99
37. ★1.00
38. ★1.00
39. ★1.00
40. Lingxi-v1.5_claude-4-sonnet-202505141.00
41. 74.601.00
42. 2025-07-201.00
43. e0.92
44. Refact.ai Agent1.00
45. 74.401.00
46. 2025-06-031.00
47. e0.98
48. Anthropic1.00
49. Tools + Claude 4 Opus (2025-05-22)0.98
50. 73.201.00
51. 2025-05-221.00
52. e0.84
53. Tools + Claude 4 Sonnet (2025-05-22)0.99
54. 72.400.92
55. 2025-05-221.00
56. e0.58
57. Qodo Command0.98
58. 71.201.00
59. 2025-07-151.00
60. Bloop1.00
61. 71.200.98
62. 2025-07-101.00
63. e0.53
64. Us!1.00
65. Warp0.99
66. 71.000.89
67. 2025-06-231.00
68. e0.65
69. Moatless Tools + Claude 4 Sonnet0.99
70. 70.801.00
71. 2025-06-111.00
72. e0.93
73. EURUPE0.94
74. TRAE1.00
75. 70.601.00
76. 2025-05-191.00
77. e0.94
78. OpenAI1.00
79. Refact.ai Agent1.00
80. 70.401.00
81. 2025-05-151.00
82. e0.91
83. OpenHands + Claude 4 Sonnet0.99
84. 70.401.00
85. 580.52
86. 2025-05-241.00
87. e0.83
88. Braintrust1.00
89. WorkOS OpenAI0.92
90. AlEngineer0.99
91. EUROPE1.00
1:14 #7 done13 line(s)

shot 7·sharpness 2529.2
1. AGENDA1.00
2. 1. Background: everything is plan and review0.98
3. 2. Balancing planning and review0.99
4. 3. Long running agents0.99
5. AIE1.00
6. 4. Let's shut my company down0.99
7. ★1.00
8. ★1.00
9. #Braintrust0.96
10. WorkOS OpenAI0.95
11. AlEngineer0.99
12. EUROPE1.00
13. 20261.00
1:45 #8 done9 line(s)

shot 8·sharpness 2420.9
1. AIE1.00
2. 1. Background: everything is plan1.00
3. ★0.99
4. ★1.00
5. and review0.99
6. AlEngineer0.97
7. AlEngineer0.97
8. EUROPE1.00
9. 20261.00
2:10 #9 done16 line(s)

shot 9·sharpness 2166.4
1. Ratios: 1843 - 20210.96
2. AIE1.00
3. ★1.00
4. ★1.00
5. Planning1.00
6. Writing code1.00
7. Reviewing1.00
8. Reviewing other1.00
9. my code1.00
10. people's code0.99
11. Source: vibes0.99
12. AlEngineer0.96
13. AlEngineer0.99
14. EUROPE1.00
15. EUROPE1.00
16. 20261.00
2:26 #10 done24 line(s)

shot 10·sharpness 2730.5
1. Ratios1.00
2. GitHub Copilot1.00
3. AIE1.00
4. 20211.00
5. Planning1.00
6. Writing code1.00
7. Reviewing1.00
8. Reviewing other1.00
9. my code0.98
10. people's code1.00
11. ★1.00
12. ★1.00
13. 20221.00
14. Planning1.00
15. Writing code1.00
16. Reviewing1.00
17. Reviewing other1.00
18. my code1.00
19. people's code1.00
20. Source: vibes0.99
21. AlEngineer0.96
22. AlEngineer0.99
23. EUROPE1.00
24. 20261.00
2:35 #11 done32 line(s)

shot 11·sharpness 3231.3
1. Stackverflow0.91
2. Ratios1.00
3. ChatGPT1.00
4. 20211.00
5. Planning1.00
6. Writing code1.00
7. Reviewing1.00
8. my code1.00
9. Reviewing other1.00
10. people's code0.99
11. AIE1.00
12. ★1.00
13. ★1.00
14. Reviewing1.00
15. Reviewing other1.00
16. 20221.00
17. Planning1.00
18. Writing code1.00
19. my code1.00
20. people's code1.00
21. 20231.00
22. Planning1.00
23. Writing code1.00
24. Reviewing my0.98
25. Reviewing other0.99
26. code1.00
27. people's code1.00
28. Source: vibes0.99
29. Engineering the future of Al1.00
30. AlEngineer0.97
31. EUROPE1.00
32. 20261.00
2:37 #12 done38 line(s)

shot 12·sharpness 3357.0
1. Ratios1.00
2. CURSOR1.00
3. Reviewing0.98
4. Reviewing other1.00
5. 20211.00
6. Planning1.00
7. Writing code0.95
8. my code1.00
9. people's code0.98
10. AIE1.00
11. 20221.00
12. Planning1.00
13. Writing code0.97
14. Reviewing1.00
15. my code0.97
16. Reviewing other0.99
17. people's code0.99
18. ★1.00
19. ★1.00
20. ★1.00
21. 20231.00
22. Planning1.00
23. Writing code0.96
24. Reviewing my0.97
25. Reviewing other0.99
26. code1.00
27. people's code1.00
28. 20241.00
29. Planning1.00
30. Writing1.00
31. Reviewing my code1.00
32. Reviewing other1.00
33. code1.00
34. people's code1.00
35. Source: vibes0.97
36. Engineering the future of Al1.00
37. AlEngineer0.99
38. 20260.99
2:46 #13 done45 line(s)

shot 13·sharpness 3700.7
1. Ratios1.00
2. 20211.00
3. Planning1.00
4. Writing code0.98
5. Reviewing1.00
6. Reviewing other1.00
7. my code1.00
8. people's code1.00
9. 20221.00
10. Planning1.00
11. Writing code0.96
12. Reviewing1.00
13. my code1.00
14. Reviewing other1.00
15. people's code1.00
16. ★1.00
17. AIE1.00
18. ★1.00
19. ★1.00
20. ★1.00
21. 20231.00
22. Planning1.00
23. Writing code0.96
24. Reviewing my1.00
25. code1.00
26. Reviewing other0.99
27. people's code1.00
28. 20241.00
29. Planning1.00
30. Writing1.00
31. Reviewing my code1.00
32. Reviewing other0.99
33. code1.00
34. people's code1.00
35. 20251.00
36. Planning1.00
37. Reviewing1.00
38. Reviewing other1.00
39. my code1.00
40. people's code1.00
41. Source: vibes0.97
42. Writing code1.00
43. Engineering the future of Al1.00
44. AlEngineer0.99
45. 20260.99
3:21 #14 done16 line(s)

shot 14·sharpness 3749.0
1. Did we get the time we used to1.00
2. spend writing code back, or did1.00
3. AIE1.00
4. it just create more work in other0.98
5. ★1.00
6. categories?1.00
7. 20251.00
8. Planning1.00
9. Reviewing1.00
10. Reviewing other1.00
11. my code1.00
12. people's code1.00
13. Writingcode1.00
14. Engineering the future of Al1.00
15. AlEngineer0.99
16. 20260.99
3:31 #15 done8 line(s)

shot 15·sharpness 2222.3
1. AIE1.00
2. 2. Balancing planning and review1.00
3. ★1.00
4. ★1.00
5. Engineering the future of Al1.00
6. AlEngineer0.99
7. EUROPE1.00
8. 20260.99
3:47 #16 done22 line(s)

shot 16·sharpness 2515.6
1. Spend a lot of time planning0.98
2. Very comprehensive plan doc0.99
3. Specs1.00
4. AIE1.00
5. ★1.00
6. Interrogation1.00
7. ★1.00
8. ★1.00
9. ★1.00
10. More time spent planning1.00
11. + Coding agent runs for longer0.99
12. +0.94
13. Less time needed for review0.98
14. Code1.00
15. Planning1.00
16. AI working1.00
17. QA1.00
18. review1.00
19. AlEngineer0.95
20. AlEngineer0.99
21. EUROPE1.00
22. 20261.00
4:36 #17 done13 line(s)

shot 17·sharpness 592.0
1. Spend a lot of time planning0.99
2. AlEngineer0.98
3. Very comprehensive plan d1.00
4. Specs1.00
5. EUROPE1.00
6. Interrogation1.00
7. More time spent planning1.00
8. Coding agent runs for lon1.00
9. Less time needed for revi1.00
10. Planning1.00
11. AI wor1.00
12. AlEngineer0.99
13. EUROPE1.00
5:07 #18 done25 line(s)

shot 18·sharpness 2666.8
1. Spend a lot of time reviewing0.99
2. Loosely defined prompts1.00
3. More time QA'ing manually1.00
4. AIE1.00
5. Synchronous1.00
6. ★1.00
7. ★1.00
8. ★1.00
9. Less time spent planning1.00
10. Coding agent yields back faster0.99
11. More back and forth1.00
12. AI1.00
13. AI1.00
14. AI1.00
15. Planning1.00
16. QA1.00
17. QA1.00
18. QA1.00
19. Code review1.00
20. working1.00
21. working1.00
22. working1.00
23. Engineering the future of Al1.00
24. AlEngineer0.98
25. 20261.00
5:26 #19 skipped

shot 19·duplicate of #18
6:19 #20 done17 line(s)

shot 20·sharpness 3163.7
1. Your 'mode' of vibe engineering1.00
2. is influenced by the type of work0.99
3. AIE1.00
4. ★1.00
5. Frontend1.00
6. Backend1.00
7. New features1.00
8. Review heavy0.98
9. Plan heavy1.00
10. Refactoring/migration1.00
11. Plan heavy0.95
12. Plan heavy1.00
13. AlEngineer0.96
14. AlEngineer0.99
15. EUROPE1.00
16. EUROPE1.00
17. 20261.00
6:37 #21 done17 line(s)

shot 21·sharpness 3175.6
1. Your 'mode' of vibe engineering1.00
2. is influenced by the type of work0.99
3. AIE1.00
4. ★1.00
5. ★1.00
6. Frontend1.00
7. Backend1.00
8. New features1.00
9. Review heavy1.00
10. Plan heavy1.00
11. Refactoring/migration1.00
12. Plan heavy1.00
13. Plan heavy1.00
14. Engineering the future of Al1.00
15. AlEngineer0.98
16. EUROPE1.00
17. 20261.00
7:14 #22 skipped

shot 22·duplicate of #12
7:57 #23 done8 line(s)

shot 23·sharpness 3166.8
1. As coding agents become more0.99
2. AIE1.00
3. capable, the time before human0.99
4. ★1.00
5. intervention is required grows1.00
6. Engineering the future of Al1.00
7. AlEngineer0.97
8. 20261.00

Transcript

232 cues· 3,294 words· 17,460 chars

chunk 0 · 0:15–0:51 · 88 words · 460 chars
0:15 Is this mic on?
0:15 Yes, this mic is on.
0:16 How are we doing?
0:19 Fucking fantastic!
0:20 Yeah!
0:21 Let's go!
0:22 All right.
0:26 So the title of this is about planning and review, but I think the real point behind this is basically what are we all going to do after AI continues to get really, really good?
0:39 I'm Louie.
0:40 I'm the founder of a startup called Vibe Kanban, and I also started the London chapter of AI Tinkerers, which is a great community if you're in London looking for events.
chunk 1 · 0:54–1:38 · 127 words · 668 chars
0:54 You should listen to me because I have done some stuff like get on the SWE bench verified leaderboard ahead of OpenAI.
1:03 This is a couple of months old now.
1:04 But anyway, you know, it's always nice to know that the people talking have done some research in the space.
1:11 The agenda for today class is we're going to, I'm gonna walk you through why I have arrived at the conclusion that basically all software engineers are gonna do all day is plan and review stuff.
1:23 And I'm gonna talk about how to think about balancing that, if that is what you do all day.
chunk 2 · 1:30–2:09 · 103 words · 554 chars
1:30 I'm gonna talk about time horizon and how agents are running for longer and how that changes the behavior of the job.
1:39 And then at the end, we're going to shut my company down.
1:42 And we'll get onto that later on.
1:45 So let's get started.
1:46 Everything is plan and review.
1:47 So work that we software engineers do.
1:51 Everybody in here is a software engineer, right?
1:53 Yes?
1:54 OK.
1:54 Most people.
1:56 We plan stuff.
1:57 We write code.
1:58 We review code.
1:59 And we review other people's code.
chunk 3 · 2:01–2:43 · 130 words · 687 chars
2:01 Roughly the work that we do.
2:04 The ratios until GitHub Copilot hit the scene were roughly this for me.
2:09 I know it depends on whether you work in a big company or a small company and things like that, but a lot of it was writing code and not very much of it compared to that was planning and reviewing code.
2:20 And what we see is over time, with things like the first version of GitHub Copilot, that basically the writing code part starts to shrink.
2:29 So, you know, chat GPT arrived, suddenly, you know, you can like generate functions and paste them in.
chunk 4 · 2:36–3:07 · 95 words · 502 chars
2:36 And then Cursor, not Cursor today, but like the original version of Cursor arrives, and then it's like able to complete a whole page of code.
2:44 And then you get Claude code, and it's like, wow, you know, I actually am not really doing much code writing anymore.
2:51 So it kind of poses an interesting question though, which is, say we were spending four hours a day coding before, does that mean I now get four hours back if I'm not doing any actual coding?
3:04 The answer of course is no, it has displaced work.
chunk 5 · 3:07–3:51 · 133 words · 718 chars
3:07 work that you or time that was previously spent doing the coding has moved it has moved to planning and reviewing i think it is an accelerant you're probably getting more done in the day but it's probably like you know you get uh 20 minutes back for every half an hour that you were you know spending coding and some other time has gone to planning and reviewing
3:30 So I want to talk about that.
3:32 What is this new way of working?
3:35 And I think there's fundamentally two approaches that people take.
chunk 6 · 3:38–4:17 · 104 words · 586 chars
3:38 I'm not going to get too specific about specs or playwright MCPs or things like that.
3:44 There's tons of fascinating talks about that at this event.
3:48 I just want to kind of conceptually ground what we're talking about here today.
3:52 The first way is the plan-based approach.
3:56 This is where you spend a lot of time upfront planning the work that you want a coding agent to do.
4:03 So the smells of whether you are doing this type of work would probably look like you're writing a very comprehensive plan doc, markdown file.
chunk 7 · 4:12–4:54 · 111 words · 657 chars
4:12 You're maybe using like one of these spec frameworks.
4:15 you're interrogating.

Chapters

0:00 Intro and agenda
1:45 Why software engineering is shifting to plan and review
3:30 The two approaches: Plan-based vs. Review-heavy
6:02 The matrix: Feature development vs. migrations and maintenance
7:27 The impact of agent execution time
9:52 Managing the 'five-minute' agent threshold
10:29 Parallelization and workflow management
12:00 Vision for future coding interfaces and 'focus maxing'
14:00 Announcement: Shutting down Vibe Kanban
18:13 Q&A: Next steps and reflections on the startup journey