read-only demo

Videos RmS5s6Wbin4

Gadgets: Personal app vibe coding that is actually safe — Kenton Varda, Cloudflare

index_state ready data_status ok

AI Engineer· published 2026-08-05· 0:18:53· en-US· indexed 2026-08-10 19:35

Open on YouTube

Scene timeline

  1. Shot 0, 0:00 to 0:03, 1 of 1 keyframes kept
  2. Shot 1, 0:03 to 0:05, 1 of 1 keyframes kept
  3. Shot 2, 0:05 to 0:12, 1 of 1 keyframes kept
  4. Shot 3, 0:12 to 0:38, 1 of 1 keyframes kept
  5. Shot 4, 0:38 to 1:03, 1 of 1 keyframes kept
  6. Shot 5, 1:03 to 1:29, 1 of 1 keyframes kept
  7. Shot 6, 1:29 to 1:55, 1 of 1 keyframes kept
  8. Shot 7, 1:55 to 2:20, 1 of 1 keyframes kept
  9. Shot 8, 2:20 to 2:46, 0 of 1 keyframes kept
  10. Shot 9, 2:46 to 3:12, 0 of 1 keyframes kept
  11. Shot 10, 3:12 to 3:37, 0 of 1 keyframes kept
  12. Shot 11, 3:37 to 4:03, 0 of 1 keyframes kept
  13. Shot 12, 4:03 to 4:29, 0 of 1 keyframes kept
  14. Shot 13, 4:29 to 4:55, 0 of 1 keyframes kept
  15. Shot 14, 4:55 to 5:20, 0 of 1 keyframes kept
  16. Shot 15, 5:20 to 5:46, 0 of 1 keyframes kept
  17. Shot 16, 5:46 to 6:12, 0 of 1 keyframes kept
  18. Shot 17, 6:12 to 6:37, 0 of 1 keyframes kept
  19. Shot 18, 6:37 to 7:03, 0 of 1 keyframes kept
  20. Shot 19, 7:03 to 7:29, 0 of 1 keyframes kept
  21. Shot 20, 7:29 to 7:54, 0 of 1 keyframes kept
  22. Shot 21, 7:54 to 7:58, 1 of 1 keyframes kept
  23. Shot 22, 7:58 to 8:26, 1 of 1 keyframes kept
  24. Shot 23, 8:26 to 8:55, 0 of 1 keyframes kept
  25. Shot 24, 8:55 to 9:24, 0 of 1 keyframes kept
  26. Shot 25, 9:24 to 9:36, 0 of 1 keyframes kept
  27. Shot 26, 9:36 to 9:38, 0 of 1 keyframes kept
  28. Shot 27, 9:38 to 9:43, 0 of 1 keyframes kept
  29. Shot 28, 9:43 to 10:12, 0 of 1 keyframes kept
  30. Shot 29, 10:12 to 10:41, 0 of 1 keyframes kept
  31. Shot 30, 10:41 to 10:46, 1 of 1 keyframes kept
  32. Shot 31, 10:46 to 11:17, 1 of 1 keyframes kept
  33. Shot 32, 11:17 to 11:38, 1 of 1 keyframes kept
  34. Shot 33, 11:38 to 11:40, 0 of 1 keyframes kept
  35. Shot 34, 11:40 to 11:43, 0 of 1 keyframes kept
  36. Shot 35, 11:43 to 12:28, 0 of 1 keyframes kept
  37. Shot 36, 12:28 to 12:30, 0 of 1 keyframes kept
  38. Shot 37, 12:30 to 12:31, 0 of 1 keyframes kept
  39. Shot 38, 12:31 to 12:34, 0 of 1 keyframes kept
  40. Shot 39, 12:34 to 13:09, 0 of 1 keyframes kept
  41. Shot 40, 13:09 to 13:45, 0 of 1 keyframes kept
  42. Shot 41, 13:45 to 14:18, 1 of 1 keyframes kept
  43. Shot 42, 14:18 to 14:51, 0 of 1 keyframes kept
  44. Shot 43, 14:51 to 15:21, 0 of 1 keyframes kept
  45. Shot 44, 15:21 to 15:52, 0 of 1 keyframes kept
  46. Shot 45, 15:52 to 16:22, 0 of 1 keyframes kept
  47. Shot 46, 16:22 to 16:52, 0 of 1 keyframes kept
  48. Shot 47, 16:52 to 17:21, 0 of 1 keyframes kept
  49. Shot 48, 17:21 to 17:51, 0 of 1 keyframes kept
  50. Shot 49, 17:51 to 18:21, 0 of 1 keyframes kept
  51. Shot 50, 18:21 to 18:24, 1 of 1 keyframes kept
  52. Shot 51, 18:24 to 18:27, 0 of 1 keyframes kept
  53. Shot 52, 18:27 to 18:37, 0 of 1 keyframes kept
  54. Shot 53, 18:37 to 18:52, 0 of 1 keyframes kept
  55. Shot 54, 18:52 to 18:53, 1 of 1 keyframes kept

55 shot(s).

keyframes kept every frame deduplicated

What was stored

cues
188
whisperx 188
chunks
32
from 188 cues
keyframes
16
kept of 55 captured
frames with text
15
301 lines read
chapters
13
from the source metadata
keyframe bytes
6.7 MB
word timings on 188 cues

Provenance

Each pipeline stage, its state and the model that produced it
stage state model started took
fetch done 2026-08-09 21:17 0s
stt done 2026-08-09 00:19 30s
chunk done 2026-08-09 00:19 0s
text_embed done 2026-08-10 19:35 0s
keyframe done 2026-08-09 00:19 2m 53s
ocr done 2026-08-09 00:22 7s
frame_embed done 2026-08-10 19:35 3s

Frames, and what the machine read

  • 0:02 #0 done2 line(s)

    shot 0·sharpness 456.7

    1. AlEngineer0.96
    2. World's Fair0.97
  • 0:03 #1 done2 line(s)

    shot 1·sharpness 663.3

    1. AlEngineer0.95
    2. World's Fair0.99
  • 0:10 #2 done24 line(s)

    shot 2·sharpness 2742.3

    1. LAB & PLATINUM SPONSORS0.99
    2. Amazon AGI Lab0.98
    3. ANTHROP\C1.00
    4. Google DeepMind1.00
    5. MINIMAX0.93
    6. OpenAI0.92
    7. Akamai1.00
    8. arize0.92
    9. aws1.00
    10. Braintrust bright data0.98
    11. B1.00
    12. Browserbase1.00
    13. docker1.00
    14. :neo4j0.90
    15. ORACLE1.00
    16. PayPal1.00
    17. qodo1.00
    18. reducto1.00
    19. Sonar1.00
    20. Makers of1.00
    21. togetherai1.00
    22. Unblocked1.00
    23. WorkOS1.00
    24. SonarQube1.00
  • 0:25 #3 done11 line(s)

    shot 3·sharpness 1886.7

    1. Jun 30 5:45 PM0.94
    2. AlEngineer0.98
    3. Personal Al codegen bre0.95
    4. World's Fair0.97
    5. PRESENTED·BY0.96
    6. Personal Al codegen0.99
    7. Microsoft1.00
    8. breaks traditional0.99
    9. cloud infrastructure.1.00
    10. Engineering the future of Al0.99
    11. World's Fair0.94
  • 1:00 #4 done13 line(s)

    shot 4·sharpness 1985.0

    1. Jun 30 5:45 PM0.97
    2. AlEngineer0.97
    3. Personal Al codegen bre0.97
    4. a3f1be641c625a33003dfb8c44dc18c8a11abe96d3?chat=0#fullscreen0.98
    5. World'sFair1.00
    6. PRESENTED·BY0.96
    7. Personal Al codegen1.00
    8. Microsoft1.00
    9. breaks traditional0.98
    10. cloud infrastructure.1.00
    11. TRACK 1· JUNE 30, 20260.95
    12. World'sFair1.00
    13. Claws & Personal Agents0.98
  • 1:11 #5 done12 line(s)

    shot 5·sharpness 1986.8

    1. Jun 30 5:46 PM0.95
    2. AlEngineer0.98
    3. Personal Al codegen bre0.96
    4. World'sFair1.00
    5. PRESENTED·BY0.96
    6. Personal Al codegen1.00
    7. Microsoft1.00
    8. breaks traditional0.99
    9. cloud infrastructure.1.00
    10. TRACK 1· JUNE 30, 20260.97
    11. World'sFair0.97
    12. Claws & Personal Agents0.99
  • 1:47 #6 done12 line(s)

    shot 6·sharpness 1968.7

    1. Jun 30 5:46 PM0.98
    2. AlEngineer0.97
    3. Personal Al codegen bre0.97
    4. World'sFair1.00
    5. PRESENTED·BY0.96
    6. Personal Al codegen0.99
    7. Microsoft1.00
    8. breaks traditional0.99
    9. cloud infrastructure.1.00
    10. TRACK 1· JUNE 30, 20260.97
    11. 1d'sFair0.98
    12. Claws & Personal Agents0.99
  • 2:12 #7 done12 line(s)

    shot 7·sharpness 1968.9

    1. Jun 30 5:47 PM0.95
    2. AlEngineer0.97
    3. Personal Al codegen bre0.97
    4. World'sFair1.00
    5. PRESENTED·BY0.97
    6. Personal Al codegen1.00
    7. Microsoft1.00
    8. breaks traditional0.99
    9. cloud infrastructure.1.00
    10. TRACK 1· JUNE 30, 20260.97
    11. 1d'sFair0.95
    12. Claws & Personal Agents0.99
  • 2:26 #8 skipped

    shot 8·duplicate of #7

  • 2:56 #9 skipped

    shot 9·duplicate of #6

  • 3:32 #10 skipped

    shot 10·duplicate of #6

  • 3:58 #11 skipped

    shot 11·duplicate of #6

  • 4:09 #12 skipped

    shot 12·duplicate of #6

  • 4:32 #13 skipped

    shot 13·duplicate of #6

  • 5:02 #14 skipped

    shot 14·duplicate of #6

  • 5:40 #15 skipped

    shot 15·duplicate of #6

  • 5:56 #16 skipped

    shot 16·duplicate of #6

  • 6:15 #17 skipped

    shot 17·duplicate of #4

  • 7:00 #18 skipped

    shot 18·duplicate of #6

  • 7:11 #19 skipped

    shot 19·duplicate of #6

  • 7:37 #20 skipped

    shot 20·duplicate of #7

  • 7:57 #21 done41 line(s)

    shot 21·sharpness 2350.1

    1. Jun 30 5:52 PM0.97
    2. AlEngineer0.99
    3. Personal Al codegen bre x0.95
    4. localhost:3000/gadget/Sa6b969af1a59b46e7da20a3f1be641c625a33003dfb8c44dc18c8a11abe96d3?chat=00.99
    5. World's Fair0.99
    6. Personal Al codegen breaks traditional cloud infrastructure0.99
    7. 50.25 ①0.85
    8. Opus 4.81.00
    9. Gadget1.00
    10. Code1.00
    11. Connections1.00
    12. Activity0.86
    13. Make a simple web page -intended to be presented as a single0.98
    14. slide while giving a conference talk which says: "Personal Al0.99
    15. codegen breaks traditional cloud infrastructure." The word0.97
    16. "breaks" should stand out between the two noun phrases.0.97
    17. PRESENTED BY0.95
    18. Ill reate a single-slide page with that statement, making "breaks" stand out.0.97
    19. Wrote 2 fles0.97
    20. client.js - server.js0.94
    21. Personal Al codegen0.98
    22. Microsoft1.00
    23. Done. Here's your conference slide.0.96
    24. It displays the line centered on a dark radial-gradient background, wth "breaks" standing0.98
    25. out between the two phrases:0.97
    26. Rendered n red with a glow0.95
    27. • Slighty tited with a jagged "crack" underine0.95
    28. breaks traditional1.00
    29. Asubtle periodic shudder animation (disabled if the viewer prefers reduced motion)0.99
    30. The text scales responsively ( clamp + viewport units). so i fils the screen whether you're0.94
    31. on a laptop or projecting full-screen. Let me know if you'd like a different color, no animation,0.97
    32. a subtitie/speaker line, or a lighe-background version.0.97
    33. cloud infrastructure.1.00
    34. Accept changes to save them to the gadget.0.95
    35. Accept changes0.94
    36. Ask a follow-up..0.99
    37. Claude Opus 4.80.99
    38. localhost:30001.00
    39. TRACK 1· JUNE 30, 20260.95
    40. World's Fair0.99
    41. Claws & Personal Agents1.00
  • 8:20 #22 done35 line(s)

    shot 22·sharpness 1676.6

    1. Jun 30 5:53 PM0.99
    2. Home-Gadgets0.99
    3. AlEngineer0.99
    4. localhost:30001.00
    5. World's Fair0.96
    6. gadgets1.00
    7. Home1.00
    8. Gatekrepers0.97
    9. Explore1.00
    10. Your gadgets1.00
    11. Gadgets0.94
    12. D0 Bluepints0.83
    13. Open a gadget or start from a saved blueprint.0.98
    14. O. Search gadgets..0.96
    15. Gadgets AlE talk0.97
    16. 0 145 ag00.81
    17. PRESENTED BY0.96
    18. Personal Al codegen breaks traditional cloud infrastructure0.99
    19. o ld ago0.78
    20. Microsoft1.00
    21. What are we working on?0.99
    22. Collaborative Whiteboard Tool0.99
    23. o 2d ago0.92
    24. Ask a question, build an app, or create an agent that works0.99
    25. Spanish Email Fiter0.98
    26. 0 2f ago0.86
    27. with your tools and data.1.00
    28. GitHub Review Tracker1.00
    29. 0 2d ago0.96
    30. Make ily ounte . S it Madimum0.63
    31. Add resource0.96
    32. GPT 5.50.97
    33. World's Fair1.00
    34. TRACK 1· JUNE 30, 20260.95
    35. Claws & Personal Agents0.99
  • 8:49 #23 skipped

    shot 23·duplicate of #22

Transcript

188 cues· 2,816 words· 14,806 chars

  1. 0:23 OK, hi.
  2. 0:25 All right, I've got a lot to talk about, so I'm going to launch right into it here.
  3. 0:29 So SWIX says that you only get to make one point at every talk, one key takeaway.
  4. 0:35 And so I figured I'd just lead with that.
  5. 0:37 My key point is personal AI code gen breaks traditional cloud infrastructure.
  6. 0:43 And to clarify what I mean about that, the word personal here is doing a lot of work.
  7. 0:49 It's load-bearing, as Claude would say.
  8. 0:53 My point is that if we want to see this future where everyone has personal apps and can personalize the apps that they run, the infrastructure we're using today for software in general is not the right thing and we need something completely different.
  9. 1:09 So to explain what I mean.
  10. 1:13 Think about the way that software is produced and distributed today.
  11. 1:18 You have a developer in an ivory tower who builds an app and then sends it down to the people, the users who use the app, and many of them are happy with it, but some of them are not.
  12. 1:31 Some of them say, this app needs some additional features for my use case.
  13. 1:37 And so they go to the developer and they say, oh, great developer, will you please grant my feature request?
  14. 1:43 Your app is literally unusable without it.
  15. 1:47 And so then the developer's representative, the product manager, takes these feature requests and files them into JIRA where they are never seen again.
  16. 1:57 But sometimes, sometimes the product manager sees the feature request and says, ah, you know, I want that too.
  17. 2:02 And then that feature request goes onto the roadmap and the developer works on it.
  18. 2:09 And the developer is implementing all these features, features that, you know, each one is only used by a small subset of users.
  19. 2:17 And each one is adding all these if statements, their code, and making things messy.
  20. 2:22 And they don't like it because the code base is becoming a mess.
  21. 2:25 And each of these features are really kind of boring to implement.
  22. 2:28 And the developer says, ah, I know what I need to do.
  23. 2:32 We need a rewrite.
  24. 2:36 We need a new architecture that has a plug-in system.
  25. 2:40 And then every one of these features can be a plugin and it can be nice and clean and easy to build and the core can stay clean.
  26. 2:48 And so the developer goes off and starts working on the new architecture with the plugin system.
  27. 2:53 And there are still feature requests coming in.
  28. 2:57 And the developer says, well, we can't do those features yet because we need the plugin system.
  29. 3:03 This will be so much easier once we have the plugin system.
  30. 3:06 And if we do it now, we're just delaying that and we'll just have to redo it later anyway.
  31. 3:12 The years go by, and the new architecture is not ready yet, and none of the features are being implemented, and people are saying, what are they doing?
  32. 3:21 This developer has given up their product, and everybody is sad.
  33. 3:27 So, AI seems to present a new alternative to this.
  34. 3:34 What if the developer could
  35. 3:38 create their app, the first version of their app, give it to the users, and the users, if they need a new feature, could say, I could ask their AI agent to write that feature just for them, add it to the app.
  36. 3:52 Then everyone gets the features they need, no one is bogged down in everyone else's features, and the developer gets to keep the core app nice and clean and beautiful.
  37. 4:03 But there's a problem with this, which is that none of the infrastructure we build software on today is remotely designed for this.
  38. 4:13 You've got Apple and Google for the past 15 years gatekeeping their systems to the point where there's five companies that can build mobile apps now because everyone else has been banned.
  39. 4:28 And it's almost easier in the United States to buy a gun than it is to get access to your own phone to install unsigned software.
  40. 4:36 You go to Google and you say, I want to install unsigned software.
  41. 4:38 And now they're going to say, oh, whoa, hold on, buddy.
  42. 4:43 You seem upset.
  43. 4:44 You should go home and think about this.
  44. 4:47 If you still want that unsigned software in 24 hours, then you can come back and talk to us.
  45. 4:54 Fortunately, we have a workaround for all of that, which is the web.
  46. 4:58 On the web, everyone can build whatever they want, and it turns out it's fine.
  47. 5:02 It's not the security disaster that Apple and Google keep telling us would happen.
  48. 5:07 So you can build whatever you want on the web, but there's a different problem on the web, which is that for the past 25 years of cloud architecture, we've been running in the wrong direction.
  49. 5:22 When you distribute a web app,
  50. 5:25 you run it on your own server.

Chapters

  1. 0:00 Personal AI codegen breaks cloud infrastructure
  2. 1:16 How feature requests die today
  3. 2:35 The plugin system rewrite trap
  4. 3:27 What if users could add their own features
  5. 5:11 Gatekeeping, and why the web is the escape hatch
  6. 7:11 Kenton Varda and Cloudflare Workers
  7. 8:39 Gadgets as an office suite, not a deploy target
  8. 9:58 Blueprints and the slide builder
  9. 11:03 One gadget per document, sharing built into the platform
  10. 12:21 Claude adds features to the app to build the slides
  11. 14:04 Why an XSS bug does not matter here
  12. 16:22 No containers, no database, running on workerd
  13. 17:24 Why it is not open source yet

Open at this second