vidtheque.

read-only demo

Videos 9fubhllmsBU

Field Guide to Fable — Thariq Shihipar, Anthropic

index_state ready data_status ok

AI Engineer· published 2026-07-06· 0:19:28· en-US· indexed 2026-08-10 19:44

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 0:19, 1 of 1 keyframes kept
  5. Shot 4, 0:19 to 0:31, 1 of 1 keyframes kept
  6. Shot 5, 0:31 to 0:42, 1 of 1 keyframes kept
  7. Shot 6, 0:42 to 0:46, 1 of 1 keyframes kept
  8. Shot 7, 0:46 to 0:50, 1 of 1 keyframes kept
  9. Shot 8, 0:50 to 0:55, 1 of 1 keyframes kept
  10. Shot 9, 0:55 to 1:35, 1 of 1 keyframes kept
  11. Shot 10, 1:35 to 1:57, 1 of 1 keyframes kept
  12. Shot 11, 1:57 to 2:08, 1 of 1 keyframes kept
  13. Shot 12, 2:08 to 2:32, 1 of 1 keyframes kept
  14. Shot 13, 2:32 to 2:37, 1 of 1 keyframes kept
  15. Shot 14, 2:37 to 3:05, 1 of 1 keyframes kept
  16. Shot 15, 3:05 to 3:38, 1 of 1 keyframes kept
  17. Shot 16, 3:38 to 4:09, 1 of 1 keyframes kept
  18. Shot 17, 4:09 to 4:21, 1 of 1 keyframes kept
  19. Shot 18, 4:21 to 4:54, 1 of 1 keyframes kept
  20. Shot 19, 4:54 to 5:26, 1 of 1 keyframes kept
  21. Shot 20, 5:26 to 5:58, 0 of 1 keyframes kept
  22. Shot 21, 5:58 to 6:10, 1 of 1 keyframes kept
  23. Shot 22, 6:10 to 6:36, 1 of 1 keyframes kept
  24. Shot 23, 6:36 to 7:02, 1 of 1 keyframes kept
  25. Shot 24, 7:02 to 7:29, 1 of 1 keyframes kept
  26. Shot 25, 7:29 to 7:57, 0 of 1 keyframes kept
  27. Shot 26, 7:57 to 8:26, 1 of 1 keyframes kept
  28. Shot 27, 8:26 to 9:03, 1 of 1 keyframes kept
  29. Shot 28, 9:03 to 9:31, 1 of 1 keyframes kept
  30. Shot 29, 9:31 to 10:00, 0 of 1 keyframes kept
  31. Shot 30, 10:00 to 10:11, 1 of 1 keyframes kept
  32. Shot 31, 10:11 to 10:47, 0 of 1 keyframes kept
  33. Shot 32, 10:47 to 10:57, 1 of 1 keyframes kept
  34. Shot 33, 10:57 to 11:24, 1 of 1 keyframes kept
  35. Shot 34, 11:24 to 11:50, 0 of 1 keyframes kept
  36. Shot 35, 11:50 to 12:27, 0 of 1 keyframes kept
  37. Shot 36, 12:27 to 12:57, 0 of 1 keyframes kept
  38. Shot 37, 12:57 to 13:34, 0 of 1 keyframes kept
  39. Shot 38, 13:34 to 13:56, 1 of 1 keyframes kept
  40. Shot 39, 13:56 to 14:13, 1 of 1 keyframes kept
  41. Shot 40, 14:13 to 14:28, 1 of 1 keyframes kept
  42. Shot 41, 14:28 to 14:44, 1 of 1 keyframes kept
  43. Shot 42, 14:44 to 15:17, 1 of 1 keyframes kept
  44. Shot 43, 15:17 to 15:48, 1 of 1 keyframes kept
  45. Shot 44, 15:48 to 16:07, 1 of 1 keyframes kept
  46. Shot 45, 16:07 to 16:29, 1 of 1 keyframes kept
  47. Shot 46, 16:29 to 17:15, 1 of 1 keyframes kept
  48. Shot 47, 17:15 to 17:40, 1 of 1 keyframes kept
  49. Shot 48, 17:40 to 18:21, 1 of 1 keyframes kept
  50. Shot 49, 18:21 to 18:57, 1 of 1 keyframes kept
  51. Shot 50, 18:57 to 19:08, 0 of 1 keyframes kept
  52. Shot 51, 19:08 to 19:11, 1 of 1 keyframes kept
  53. Shot 52, 19:11 to 19:27, 0 of 1 keyframes kept

53 shot(s).

keyframes kept every frame deduplicated

What was stored

cues
218
whisperx 218
chunks
34
from 218 cues
keyframes
43
kept of 53 captured
frames with text
43
1,245 lines read
chapters
5
from the source metadata
keyframe bytes
6.2 MB
word timings on 218 cues

Provenance

Each pipeline stage, its state and the model that produced it
stage state model started took
fetch done — 2026-08-10 00:57 1m 22s
stt done — 2026-08-10 00:58 21s
chunk done — 2026-08-10 00:58 0s
text_embed done — 2026-08-10 19:44 0s
keyframe done — 2026-08-10 00:58 2m 06s
ocr done — 2026-08-10 01:00 18s
frame_embed done — 2026-08-10 19:44 8s

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 2736.8

    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 data0.98
    11. B1.00
    12. Browserbase1.00
    13. docker1.00
    14. :neo4j0.93
    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

  • 0:17 #3 done2 line(s)

    shot 3·sharpness 237.2

    1. ANTHROP\C1.00
    2. AIE0.97

  • 0:30 #4 done87 line(s)

    shot 4·sharpness 1608.6

    1. World'sFair0.99
    2. Google DeepMind1.00
    3. World'sFair1.00
    4. :neo4j0.88
    5. World'sFai0.99
    6. Z.AI0.99
    7. Id's Fair0.94
    8. vast.ai1.00
    9. World'sFai0.97
    10. Ref.1.00
    11. World'sFa0.99
    12. RedHat1.00
    13. World'sFai0.99
    14. m1.00
    15. API0.97
    16. World'sFai0.99
    17. Modal1.00
    18. World'sFai1.00
    19. promptql1.00
    20. World'sFa1.00
    21. THEVELOCITYROOM1.00
    22. Id'sFair0.97
    23. AIEng0.85
    24. jineer0.99
    25. anica1.00
    26. WorIOTHARIQ SHIHIPAR0.91
    27. W0.99
    28. Id's Fair0.92
    29. I'sFair0.99
    30. MERGE1.00
    31. Id'sFair0.95
    32. twilio0.93
    33. W1.00
    34. Id's Fair0.89
    35. 0O INNGEST0.86
    36. World'sFai0.99
    37. BAND1.00
    38. F0.93
    39. 70.99
    40. World's Fair0.97
    41. Cleric1.00
    42. World'sFai0.99
    43. AA0.65
    44. Snorkel0.97
    45. World'sFai0.99
    46. Z.AI0.98
    47. World'sFa1.00
    48. MEMBER OF TECHNICAL STAFF0.99
    49. qodo0.99
    50. World'sFai0.95
    51. PayPal1.00
    52. Id'sFair0.92
    53. :neo4j0.94
    54. World'sFai0.99
    55. Google DeepMind0.99
    56. World'sFal0.94
    57. arize1.00
    58. World'sFa0.94
    59. ANTHROP\C1.00
    60. WorkOS0.93
    61. World'sFai1.00
    62. Amazon AGI Lab0.97
    63. World's Fai0.96
    64. Microsoft0.97
    65. World'sFa0.97
    66. ORACLE1.00
    67. Id'sFai0.93
    68. Braintrust1.00
    69. World'sFai0.98
    70. OpenAl0.95
    71. World'sF0.99
    72. MINIMAX0.99
    73. World'sFa0.95
    74. aws0.85
    75. World'sFal0.94
    76. togetherai1.00
    77. World'sFai0.96
    78. Id's Fain0.91
    79. DATADOG1.00
    80. Res1.00
    81. LanceDB1.00
    82. World'sFair0.89
    83. World'sFai0.96
    84. Id'sFair0.94
    85. Microsoft1.00
    86. cognee1.00
    87. Ben0.91

  • 0:34 #5 done75 line(s)

    shot 5·sharpness 3397.4

    1. World's Fair0.97
    2. wor0.99
    3. Fair1.00
    4. THIE VELOCITY ROOM0.95
    5. mezmo0.99
    6. briga tats0.68
    7. io0.84
    8. fiddler0.93
    9. air1.00
    10. INNGEST1.00
    11. Oaste0.54
    12. el0.75
    13. World's Fair0.98
    14. ROK0.69
    15. World's Fair0.97
    16. BAND1.00
    17. AUTOMATTIC1.00
    18. air0.89
    19. Z.AI0.97
    20. World's Fain0.93
    21. Keycard1.00
    22. :neo4j0.95
    23. World'sFai1.00
    24. Cleric1.00
    25. Meticulous0.92
    26. World's Fair0.95
    27. qodo0.99
    28. World'sFal0.94
    29. GGRAMITEE0.95
    30. S0.71
    31. World's Fair0.99
    32. Amazon AGI Lab1.00
    33. Google DeepMind0.97
    34. World's Fa0.92
    35. World's Fai0.94
    36. World's Fai0.92
    37. arize1.00
    38. PayPal0.96
    39. rid'sF0.82
    40. ASSIAN1.00
    41. Gradium0.94
    42. Danid0.54
    43. Braintrust1.00
    44. World's Fair0.97
    45. Microsoft1.00
    46. World's Fa0.87
    47. cto1.00
    48. ANTHROPC0.97
    49. OpenAl0.97
    50. OF0.74
    51. World's Fair0.97
    52. World's Fai0.97
    53. MINIMAX1.00
    54. bright data0.97
    55. World'sFair1.00
    56. Akamai1.00
    57. World'sFa1.00
    58. ANTHROPIC1.00
    59. World'sFair1.00
    60. DATADOG1.00
    61. World'sFa1.00
    62. Resolve.ai1.00
    63. id'sFair0.75
    64. Aurbyte0.84
    65. Ogeptle0.70
    66. DERI0.91
    67. Optiver0.94
    68. World's Fair0.96
    69. builder.io1.00
    70. World's Fai0.91
    71. Ravenna1.00
    72. World'sFa1.00
    73. d'sFa0.92
    74. World's Fair0.96
    75. World'sFa0.96

  • 0:45 #6 done13 line(s)

    shot 6·sharpness 466.6

    1. ir0.98
    2. AmazonAo.L50.82
    3. World'sFa1.00
    4. AlEngineer0.98
    5. OpenAI0.95
    6. st1.00
    7. World'sFa1.00
    8. AlEngineer0.98
    9. togeth1.00
    10. World'sFa1.00
    11. AIEng..0.87
    12. World's1.00
    13. DATADO1.00

  • 0:48 #7 done157 line(s)

    shot 7·sharpness 3126.3

    1. World'sFair0.94
    2. ORACLE1.00
    3. World's Fair0.98
    4. arize1.00
    5. World's Fair0.96
    6. Google DeepMind1.00
    7. World's Fair0.96
    8. :neo4]0.90
    9. World's Fair0.94
    10. Z.AI1.00
    11. World's Fair0.94
    12. bright data1.00
    13. World's Fair0.95
    14. paper compute co.0.98
    15. World's Fal0.96
    16. extend1.00
    17. World's Fa0.95
    18. vast.ai1.00
    19. World's Fair0.94
    20. Ref.1.00
    21. World's Fair0.98
    22. RedHat1.00
    23. World's Fa0.95
    24. mezmo*0.94
    25. World'sFa0.96
    26. stigg1.00
    27. World's Fa0.97
    28. World's Fair0.96
    29. RELAI0.99
    30. World'sFai0.97
    31. World's Fai0.96
    32. Modal1.00
    33. World's Fair0.93
    34. promptql1.00
    35. World'sFal0.93
    36. THE VELOCITYROOM0.98
    37. World's Fair0.91
    38. fiddler1.00
    39. World'sFal0.94
    40. SurrealDe0.84
    41. World'sFal0.97
    42. ZERO0.99
    43. World's Fai0.96
    44. AlEngineer0.97
    45. comet1.00
    46. World'sFal0.96
    47. authe0.97
    48. World'sFa0.95
    49. World's Fal0.93
    50. SOIOJIO0.87
    51. World'sFa0.99
    52. granica1.00
    53. World'sFal0.96
    54. World's Fal0.94
    55. PRIOR1.00
    56. World's Fair1.00
    57. DigitalOcean1.00
    58. World'sFal0.92
    59. World's Fai0.88
    60. POSTMAN0.97
    61. World'sFa0.94
    62. Composio0.97
    63. World'sFal0.93
    64. World'sFai0.99
    65. Modular1.00
    66. World's Fai0.94
    67. MERGE1.00
    68. World's Fa0.96
    69. AUTOMATTIC1.00
    70. World's Fal0.92
    71. Buildkite1.00
    72. yugabyteDB1.00
    73. World'sFal0.96
    74. Zed1.00
    75. World'sFa0.95
    76. Keycard1.00
    77. Meticulous1.00
    78. World's Fal0.88
    79. World'sFal0.97
    80. Daytona1.00
    81. World'sFair0.88
    82. twillo0.94
    83. World's Fain0.90
    84. GRAVITEE1.00
    85. World's Fal0.95
    86. PlanetScale0.99
    87. Temporal1.00
    88. World's Fal0.93
    89. Llamalndex0.98
    90. World's Fair0.87
    91. 0O INNGEST0.85
    92. World's Fair0.98
    93. BAND1.00
    94. World'sFair1.00
    95. Cleric1.00
    96. World'sFair0.98
    97. À ATLASSIAN0.96
    98. FACTORY1.00
    99. baseten1.00
    100. Snorkel0.98
    101. World'sFa1.00
    102. Z.AI0.99
    103. World'sFa1.00
    104. qodo0.98
    105. PayPal1.00
    106. Gradium1.00
    107. ANTHROPIC0.98
    108. Amazon AGI Lab0.99
    109. ;neo4j0.87
    110. World'sFal0.94
    111. Google DeepMind1.00
    112. World'sFai0.94
    113. arize1.00
    114. World'sFa1.00
    115. reducto1.00
    116. Microsoft1.00
    117. OpenAl0.96
    118. Vorld'sFa0.98
    119. WorkOS0.99
    120. Amazon AGi0.96
    121. Microsoft0.99
    122. ORACLE1.00
    123. bright data0.98
    124. Microsoft1.00
    125. Braintrust1.00
    126. OpenAl0.94
    127. MINIMAX1.00
    128. Z.AI0.97
    129. Vorld'sFa0.93
    130. ANTHROPIC0.98
    131. World'sF0.95
    132. snyk1.00
    133. World'sFal0.95
    134. World'sFa0.96
    135. Lightrun1.00
    136. forld'sFai0.93
    137. DATADOG1.00
    138. Vorld'sFa1.00
    139. Resolve.ai1.00
    140. World's Fal0.82
    141. Vorld'sF0.95
    142. LanceDB1.00
    143. World'sFa1.00
    144. builderic0.94
    145. World's Fa0.95
    146. Ravenna1.00
    147. World'sFal0.95
    148. CopilotKit1.00
    149. descupe0.90
    150. World'sFal0.97
    151. TOPK1.00
    152. World's Fair0.94
    153. co0.76
    154. 2e0.78
    155. Fal0.92
    156. eNCORD0.92
    157. Won.0.83

  • 0:53 #8 done11 line(s)

    shot 8·sharpness 434.7

    1. Amazon AGI Lab0.99
    2. World's Fair0.99
    3. AIEngineer0.97
    4. OpenAl0.94
    5. World'sFair1.00
    6. AlEngineer0.99
    7. togethe1.00
    8. World's Fair0.97
    9. AIEngineer0.96
    10. DATADOG1.00
    11. World's Fa0.99

  • 1:27 #9 done11 line(s)

    shot 9·sharpness 442.7

    1. nazon AGI Lab0.97
    2. World'sFair1.00
    3. AlEngineer0.97
    4. OpenAI0.92
    5. World's Fair0.96
    6. ineer1.00
    7. together.ai0.99
    8. sFair1.00
    9. AIEngineer0.97
    10. World'sFair1.00
    11. ATADOG1.00

  • 1:54 #10 done20 line(s)

    shot 10·sharpness 2059.8

    1. AlEngineer1.00
    2. World'sFair1.00
    3. △△0.95
    4. Id's Fair0.94
    5. IVIIC0.96
    6. The Map is Opening Up0.94
    7. AlEngi0.98
    8. enAl0.99
    9. World'0.97
    10. 3/401.00
    11. Engineer -0.95
    12. Id'sF0.96
    13. Aka0.99
    14. In the Land of Al Agents, the Verifiers Are King1.00
    15. AlEngi0.92
    16. Tariq Shaukat / Chief Executive Officer0.98
    17. Sonar1.00
    18. Makers of0.99
    19. ATADO1.00
    20. World'0.99

  • 2:02 #11 done23 line(s)

    shot 11·sharpness 2308.4

    1. AlEngineer1.00
    2. World'sFair1.00
    3. PRESENTED BY0.99
    4. AField Guide0.97
    5. Microsoft1.00
    6. to Fable0.98
    7. AGILaD0.96
    8. World'sF0.98
    9. AI0.89
    10. Fair1.00
    11. eer0.84
    12. Open/0.85
    13. 4/400.99
    14. ngineer-0.99
    15. ther1.00
    16. d'sF0.99
    17. In the Land of Al Agents, the Verifiers Are King1.00
    18. Fair1.00
    19. ser0.79
    20. DATAD1.00
    21. Tariq Shaukat / Chief Executive Officer0.98
    22. Sonar1.00
    23. Makers of0.99

  • 2:24 #12 done29 line(s)

    shot 12·sharpness 3163.0

    1. AlEngineer0.96
    2. World'sFair1.00
    3. 11.00
    4. Unhobbling Claude0.99
    5. AField1.00
    6. PRESENTED BY0.99
    7. 21.00
    8. Findingyour1.00
    9. Microsoft1.00
    10. Guide1.00
    11. Unknowns1.00
    12. toFable:1.00
    13. 31.00
    14. Dealing with the Grief0.99
    15. Being Unreasonable0.97
    16. GI LaD0.88
    17. World'sFa1.00
    18. Fair1.00
    19. OpenA1.00
    20. 5/401.00
    21. neer1.00
    22. nerai0.98
    23. sFa1.00
    24. In the Land of Al Agents, the Verifiers Are King1.00
    25. Fair1.00
    26. ATADC1.00
    27. Tariq Shaukat / Chief Executive Officer0.98
    28. Sonar1.00
    29. Makers of0.99

  • 2:34 #13 done22 line(s)

    shot 13·sharpness 1756.6

    1. AlEngineer0.98
    2. World'sFair1.00
    3. Part11.00
    4. PRESENTED BY0.99
    5. Unhobbling1.00
    6. Microsoft1.00
    7. Claude1.00
    8. orid'stair|0.84
    9. AI0.93
    10. )penAl0.91
    11. Worl1.00
    12. 6/400.99
    13. AlEngi1.00
    14. orld1.00
    15. A1.00
    16. In the Land of Al Agents, the Verifiers Are King0.99
    17. DATAL0.99
    18. Worl1.00
    19. Al0.86
    20. Tariq Shaukat / Chief Executive Officer0.98
    21. Sonar1.00
    22. Makers of0.99

  • 2:54 #14 done22 line(s)

    shot 14·sharpness 2350.6

    1. AlEngineer0.99
    2. World'sFair1.00
    3. PRESENTED BY1.00
    4. Models are grown,0.97
    5. Microsoft1.00
    6. not designed.1.00
    7. IAGILAD0.94
    8. worid's0.93
    9. igineer0.97
    10. I'sFair0.99
    11. Oper1.00
    12. 7/401.00
    13. IEngine0.95
    14. jett arc0.84
    15. Id's0.89
    16. In the Land of Al Agents, the Verifiers Are King1.00
    17. I'sFair0.99
    18. gineer1.00
    19. ,DATA0.92
    20. Tariq Shaukat / Chief Executive Officer0.98
    21. Sonar1.00
    22. Makers of0.99

  • 3:34 #15 done23 line(s)

    shot 15·sharpness 2947.2

    1. AlEngineer0.97
    2. World's Fair0.96
    3. What contains them is us — the0.98
    4. PRESENTED BY1.00
    5. Microsoft1.00
    6. harness we put them in, and the way0.98
    7. we prompt them.0.96
    8. :UIIAGILdD0.76
    9. wori0.92
    10. - AlEngineer0.98
    11. rld's Fair1.00
    12. Op1.00
    13. 8/401.00
    14. All1.00
    15. togeth0.91
    16. Norl1.00
    17. In the Land of Al Agents, the Verifiers Are King0.99
    18. rld's Fa0.96
    19. - AlEngineer0.98
    20. ED0.61
    21. Tariq Shaukat / Chief Executive Officer0.99
    22. Sonar1.00
    23. Makers of0.99

  • 3:48 #16 done37 line(s)

    shot 16·sharpness 3106.7

    1. AlEngineer1.00
    2. Broski1.00
    3. World'sFair1.00
    4. @broskiFGC1.00
    5. its cool how half the global economy is contingent on this0.99
    6. pokemon ending with aw1.00
    7. There is currently only one official Pokémon whose name ends with the letters "aw":1.00
    8. PRESENTED.BY0.98
    9. The Pokémon that0.98
    10. Oshawott (Wait, no, Oshawott ends with "ott").0.97
    11. Microsoft1.00
    12. endin“aw"0.96
    13. Let's double-check the entire National Pokédex: there is actually no official Pokémon0.99
    14. whose English name ends with "aw".1.00
    15. If you are thinking of specific Pokémon characters, moves, or adjacent terms, you might0.99
    16. be looking for:1.00
    17. • Moves: Claw Sharpen or Metal Claw0.98
    18. JIIAUILaD0.68
    19. vvoric0.64
    20. • Abilities: Rough Skin (not ending in aw)0.98
    21. • Pre-evolutions/Names containing "claw": Pokémon like Anorith, Armaldo.0.98
    22. Clawitzer, or Snease! (the Sharp Claw Pokémon), but none of their actual names end0.99
    23. rld's Fai.0.95
    24. AIEngineer0.97
    25. Ope0.83
    26. in "aw".0.90
    27. 9/400.99
    28. AlEn0.93
    29. ogethe0.96
    30. Norlc0.89
    31. In the Land of Al Agents, the Verifiers Are King1.00
    32. rld's Fa0.94
    33. AlEngineer0.99
    34. DA1.00
    35. TariqShaukat / Chief Executive Officer0.98
    36. Sonar1.00
    37. Makers of0.99

  • 4:18 #17 done40 line(s)

    shot 17·sharpness 2798.6

    1. AlEngineer0.99
    2. World's Fair0.97
    3. Opus 4.7 (1M context) with xhigh effort . Claude Max0.98
    4. Claude Code v2.1.1430.99
    5. ~/src/exec1.00
    6. /clear1.00
    7. Fetchevery1.00
    8. (no content)1.00
    9. get a list of all pokemon then filter it to ones ending in 'aw'0.99
    10. PRESENTED BY0.99
    11. Bash(curl -s "https://pokeapi.co/api/v2/pokemon?limit=2000" I python3 -c "import json,sys; data0.98
    12. Pokémon. Write a0.98
    13. data['results']]; pri..)0.97
    14. Microsoft1.00
    15. Total: 13500.99
    16. Ending in aw: 21.00
    17. script. Filter for1.00
    18. croconaw1.00
    19. drednaw1.00
    20. Out of 1350 Pokemon, 2 end in "aw":0.98
    21. 661.00
    22. aw."0.99
    23. 990.87
    24. croconaw1.00
    25. drednaw1.00
    26. Churned for 11s1.00
    27. LUIIAGILdD0.84
    28. wor0.71
    29. bypass permissions on (shift+tab to cycle)0.99
    30. - AlEngineer0.94
    31. orld's Faii0.94
    32. Op1.00
    33. 10/401.00
    34. AI0.83
    35. togeth0.97
    36. lorl0.93
    37. In the Land of Al Agents, the Verifiers Are King1.00
    38. orld's Fa0.97
    39. - AlEngineer0.95
    40. D1.00

  • 4:28 #18 done

    shot 18·sharpness 2578.2

  • 4:57 #19 done

    shot 19·sharpness 2674.5

  • 5:30 #20 skipped

    shot 20·duplicate of #19

  • 6:00 #21 done

    shot 21·sharpness 2791.1

  • 6:18 #22 done

    shot 22·sharpness 2991.1

  • 6:49 #23 done

    shot 23·sharpness 2150.2

The page's on-screen-text budget of 600 lines is spent, so the last cards in this grid list fewer lines than they hold. Narrow the page with ?frames= to read them.

Transcript

218 cues· 3,181 words· 16,148 chars

  1. 0:12 Please welcome to the stage, member of technical staff at Anthropic, Tariq Shihapar.
  2. 0:33 Hi, everyone.
  3. 0:33 I'm Tarek.
  4. 0:34 I work at Anthropic on Cloud Code.
  5. 0:37 Before we get started, we have a tradition on Cloud Code where we take a selfie before a talk.
  6. 0:41 So if you don't mind, if you strike a pose with me, I'll take a quick selfie at AI Engineer.
  7. 0:50 OK, incredible.
  8. 0:51 Well, yeah, to kick things off, like we said, Fable is back.
  9. 1:00 We're rolling it out later today.
  10. 1:02 Stay tuned for exact timeline.
  11. 1:05 Me and Kat Wu and Simon Wilson will be doing a fireside chat at 12.30.
  12. 1:10 We might have some updates for you then.
  13. 1:14 But Fable is a model I'm just so, so excited about.
  14. 1:18 It's one of those anthropic models where you're just gonna remember it.
  15. 1:22 Like, Sonnet 3.5 new, Opus 4, Opus 4.5.
  16. 1:27 It's a model that I just have a lot of affection and excitement for.
  17. 1:31 And the best way to describe Fable to me is the map is opening up.
  18. 1:37 You are playing an RPG and you've been on the tutorial
  19. 1:42 And now you get to the point where the open world starts.
  20. 1:46 And there's so much that you can do and explore.
  21. 1:50 But it's also a little bit intimidating and confusing, because there's so much you can do.
  22. 1:56 And so what I wanted to do in this talk is give you guys a field guide to Fable.
  23. 2:03 How do you work with this new class of models?
  24. 2:07 So I've got four parts to it.
  25. 2:10 I've been working on this as a series of articles and blog posts.
  26. 2:14 But when we announced Fable was coming out, I was like, OK, let me do all of this at once, at the talk, speed run.
  27. 2:24 So the four parts, unhobbling Claude, finding your unknowns, dealing with the grief, and being unreasonable.
  28. 2:33 So first, unhobbling Claude.
  29. 2:38 I think something we say really often is that the models are grown, not designed.
  30. 2:44 We don't wake up and be like, we need 99% on SweBench.
  31. 2:50 The models are something we grow carefully.
  32. 2:52 We give it data and feedback and compute.
  33. 2:56 But ultimately, it's something that we, it's a little bit organic, and we sort of figure out and learn with the model as we use it.
  34. 3:06 And so what that also means is that what contains them is us.
  35. 3:11 The harness we put them in and the way we prompt them is basically a function of our understanding of Claude.
  36. 3:19 And by unhobbling it, I mean,
  37. 3:21 how can we understand Claude better to unleash it?
  38. 3:25 And we need to understand Fable more.
  39. 3:28 So I think one of my points is that we're still so early, and I think there's a lot more understanding in Fable to unlock.
  40. 3:39 And I think I'll give you a quick example about how models get smarter, because it's a little bit unintuitive.
  41. 3:46 I saw this viral tweet a couple of weeks ago being like, why can't LLMs say which Pokemon end in AW?
  42. 3:54 There are a thousand Pokemon, right, and turns out there are two whose names end in A-W, Croconaw and Dreadnaw, right, and it turns out if you ask like a normal chat model, it can't answer it, which is kind of confusing because like, you know, it definitely knows all the names of the Pokemon, right, but if you ask Cloud Code, it can, right, because what it does is that it fetches every Pokemon and writes a script to filter for A-W, right, and so,
  43. 4:22 This is what I mean by unhobbling Claude.
  44. 4:26 We call this capability overhang.
  45. 4:28 Claude gets smarter in spiky ways.
  46. 4:31 So it doesn't just remember every Pokemon and reason through it, but if you give it the code execution tool, it can find the two Pokemon that end with AW.
  47. 4:41 And so this is, I think, part of the challenge with Fable is figuring out this capability overhang.
  48. 4:46 What is now possible?
  49. 4:47 And I think this is a discovery that I'm excited to go on with you.
  50. 4:51 To make this a little bit clearer, I'm going to talk about a few different examples of how models have progressed in the past.

Chapters

  1. 0:00 Introduction and setting the stage for Fable
  2. 2:32 Unhobbling Claude: Understanding model behavior
  3. 9:08 Finding your unknowns: Navigating the gap between map and territory
  4. 14:29 Reflecting on the emotional shift in coding productivity
  5. 16:30 Being unreasonable: Demanding good, fast, and cheap results

Open at this second