vidtheque.

read-only demo

Videos RGSFUqzqErE

On AI and Knowledge — Pablo Castro, Distinguished Engineer & CVP for AI Knowledge, Microsoft

index_state ready data_status ok

AI Engineer· published 2026-07-17· 0:17:35· en-US· indexed 2026-08-11 02:40

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 0:20, 1 of 1 keyframes kept
  5. Shot 4, 0:20 to 0:31, 1 of 1 keyframes kept
  6. Shot 5, 0:31 to 0:35, 1 of 1 keyframes kept
  7. Shot 6, 0:35 to 1:04, 1 of 1 keyframes kept
  8. Shot 7, 1:04 to 1:11, 1 of 1 keyframes kept
  9. Shot 8, 1:11 to 1:17, 1 of 1 keyframes kept
  10. Shot 9, 1:17 to 1:51, 1 of 1 keyframes kept
  11. Shot 10, 1:51 to 2:18, 1 of 1 keyframes kept
  12. Shot 11, 2:18 to 2:46, 1 of 1 keyframes kept
  13. Shot 12, 2:46 to 3:13, 1 of 1 keyframes kept
  14. Shot 13, 3:13 to 3:41, 0 of 1 keyframes kept
  15. Shot 14, 3:41 to 4:08, 1 of 1 keyframes kept
  16. Shot 15, 4:08 to 4:35, 1 of 1 keyframes kept
  17. Shot 16, 4:35 to 5:11, 1 of 1 keyframes kept
  18. Shot 17, 5:11 to 5:41, 1 of 1 keyframes kept
  19. Shot 18, 5:41 to 6:12, 1 of 1 keyframes kept
  20. Shot 19, 6:12 to 6:43, 1 of 1 keyframes kept
  21. Shot 20, 6:43 to 7:13, 1 of 1 keyframes kept
  22. Shot 21, 7:13 to 7:38, 1 of 1 keyframes kept
  23. Shot 22, 7:38 to 8:06, 1 of 1 keyframes kept
  24. Shot 23, 8:06 to 8:34, 1 of 1 keyframes kept
  25. Shot 24, 8:34 to 9:02, 1 of 1 keyframes kept
  26. Shot 25, 9:02 to 9:30, 0 of 1 keyframes kept
  27. Shot 26, 9:30 to 9:58, 1 of 1 keyframes kept
  28. Shot 27, 9:58 to 10:16, 1 of 1 keyframes kept
  29. Shot 28, 10:16 to 10:59, 1 of 1 keyframes kept
  30. Shot 29, 10:59 to 11:10, 1 of 1 keyframes kept
  31. Shot 30, 11:10 to 11:13, 1 of 1 keyframes kept
  32. Shot 31, 11:13 to 11:50, 0 of 1 keyframes kept
  33. Shot 32, 11:50 to 12:16, 1 of 1 keyframes kept
  34. Shot 33, 12:16 to 12:41, 1 of 1 keyframes kept
  35. Shot 34, 12:41 to 13:03, 1 of 1 keyframes kept
  36. Shot 35, 13:03 to 13:07, 1 of 1 keyframes kept
  37. Shot 36, 13:07 to 13:44, 1 of 1 keyframes kept
  38. Shot 37, 13:44 to 14:21, 1 of 1 keyframes kept
  39. Shot 38, 14:21 to 14:39, 0 of 1 keyframes kept
  40. Shot 39, 14:39 to 15:10, 1 of 1 keyframes kept
  41. Shot 40, 15:10 to 15:41, 0 of 1 keyframes kept
  42. Shot 41, 15:41 to 16:11, 1 of 1 keyframes kept
  43. Shot 42, 16:11 to 16:42, 0 of 1 keyframes kept
  44. Shot 43, 16:42 to 16:57, 1 of 1 keyframes kept
  45. Shot 44, 16:57 to 17:13, 1 of 1 keyframes kept
  46. Shot 45, 17:13 to 17:17, 0 of 1 keyframes kept
  47. Shot 46, 17:17 to 17:34, 0 of 1 keyframes kept

47 shot(s).

keyframes kept every frame deduplicated

What was stored

cues
133
whisperx 133
chunks
30
from 133 cues
keyframes
39
kept of 47 captured
frames with text
39
1,728 lines read
chapters
10
from the source metadata
keyframe bytes
6.3 MB
word timings on 133 cues

Provenance

Each pipeline stage, its state and the model that produced it
stage state model started took
fetch done — 2026-08-11 02:35 1m 34s
stt done — 2026-08-11 02:37 19s
chunk done — 2026-08-11 02:37 0s
text_embed done — 2026-08-11 02:37 13s
keyframe done — 2026-08-11 02:37 1m 58s
ocr done — 2026-08-11 02:39 24s
frame_embed done — 2026-08-11 02:40 7s

Frames, and what the machine read

  • 0:02 #0 done2 line(s)

    shot 0·sharpness 454.1

    1. AlEngineer0.95
    2. World's Fair0.97

  • 0:03 #1 done2 line(s)

    shot 1·sharpness 665.2

    1. AIEngineer0.95
    2. World's Fair0.99

  • 0:10 #2 done24 line(s)

    shot 2·sharpness 2746.2

    1. LAB & PLATINUM SPONSORS0.99
    2. Amazon AGI Lab0.98
    3. ANTHROP\C1.00
    4. Google DeepMind1.00
    5. MINIMAX0.94
    6. OpenAI0.92
    7. Akamai1.00
    8. arize0.93
    9. aws1.00
    10. Braintrust bright data0.98
    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

  • 0:18 #3 done2 line(s)

    shot 3·sharpness 198.3

    1. Microsoft1.00
    2. AIE0.98

  • 0:22 #4 done8 line(s)

    shot 4·sharpness 866.1

    1. PABLOCASTRO1.00
    2. iniaial0.57
    3. CVP & DISTINGUISHED ENGINEER1.00
    4. Microsoft0.94
    5. AIE1.00
    6. iln10.63
    7. World's Fair0.98
    8. Microsoft1.00

  • 0:34 #5 done186 line(s)

    shot 5·sharpness 3214.2

    1. ORACLE1.00
    2. World's Fair0.99
    3. arize1.00
    4. World's Fair0.94
    5. Google DeepMind0.99
    6. World's Fair0.95
    7. :neo4j0.88
    8. World's Fair0.99
    9. Z.AI0.99
    10. World's Fair0.97
    11. bright data0.99
    12. World's Fair0.97
    13. Browserbo1.00
    14. co.0.67
    15. World's Fair0.99
    16. extend1.00
    17. World's Fair0.96
    18. vast.ai1.00
    19. World's Fair0.98
    20. Ref.1.00
    21. World's Fair0.97
    22. RedHat1.00
    23. World's Fair0.94
    24. mezmo+0.94
    25. World's Fair0.93
    26. stigg1.00
    27. World's Fai0.97
    28. RELAI0.97
    29. World's Fair0.92
    30. VAPI1.00
    31. World's Fair0.98
    32. Modal1.00
    33. World's Fair0.98
    34. promptql1.00
    35. World's Fair0.97
    36. THE VELOCITY ROOM0.99
    37. World's Fair0.98
    38. fiddler1.00
    39. World's Fair0.93
    40. Supercondue0.95
    41. World's Fair0.98
    42. ZERO0.94
    43. World's Fair0.92
    44. AIEngineer0.96
    45. comet1.00
    46. World's Fair0.96
    47. authe0.99
    48. World's Fai0.94
    49. SOIO.IO0.79
    50. World's Fair0.96
    51. granica1.00
    52. World's Fair0.99
    53. dash01.00
    54. World's Fair0.99
    55. PRIOR1.00
    56. World's Fair1.00
    57. an1.00
    58. World's Fair0.99
    59. Vemce0.91
    60. World's Fair0.95
    61. POSTMAN1.00
    62. World's Fair0.99
    63. €Composio0.95
    64. World's Fai0.91
    65. Modular1.00
    66. World's Fair0.99
    67. MERGE1.00
    68. World's Fair0.98
    69. AUTOMATTIC1.00
    70. World's Fair0.99
    71. Buildki1.00
    72. World's Fair0.99
    73. Zed0.99
    74. World's Fair0.94
    75. Keycard1.00
    76. World's Fair0.98
    77. Meticulous1.00
    78. World's Fai0.93
    79. Daytona1.00
    80. World's Fair0.92
    81. twilio0.99
    82. World's Fair0.98
    83. G GRAVITEE0.89
    84. World's Fair0.95
    85. PlanetSca0.96
    86. World's Far0.98
    87. w Llamalndex0.97
    88. World's Fair0.99
    89. INNGEST0.96
    90. World's Fair0.98
    91. BAND1.00
    92. World's Fair0.99
    93. Cleric1.00
    94. World's Fair0.99
    95. A ATLASSIAN0.94
    96. World's Fair0.95
    97. FACTORY1.00
    98. World's Fai0.97
    99. baseten1.00
    100. World's Fair0.97
    101. Snorkel1.00
    102. World's Fair0.99
    103. Z.AI0.99
    104. World's Fair0.99
    105. qodo0.98
    106. World's Fair0.96
    107. PayPal1.00
    108. World's Fair0.98
    109. Gradium1.00
    110. World's Fair0.95
    111. ANTHROP1.00
    112. b1.00
    113. World's Fair0.99
    114. Browserbase1.00
    115. World's Fair0.97
    116. :neo4j0.94
    117. World's Fair0.97
    118. Google DeepMind0.99
    119. World's Fair0.98
    120. arize1.00
    121. World's Fair1.00
    122. reducto1.00
    123. World's Fair0.97
    124. Microsoft1.00
    125. World's Fai0.88
    126. OpenAl0.97
    127. World's Fair0.96
    128. WorkOs0.91
    129. World's Fair0.97
    130. Amazon AGI Lab0.99
    131. World's Fai-0.92
    132. Microsoft1.00
    133. World's Fair0.99
    134. ORACLE1.00
    135. World's Fair0.99
    136. bright data0.99
    137. World's Fair0.98
    138. Google DeepM1.00
    139. w0.89
    140. World's Fair0.91
    141. Braintrust1.00
    142. World's Fair0.96
    143. OpenA1.00
    144. World's Fair0.97
    145. MINIMAX1.00
    146. World's Fair0.95
    147. Z.AI0.98
    148. World's Fair0.94
    149. ANTHROPIC0.99
    150. World'sFai0.95
    151. k1.00
    152. World's Far0.98
    153. aws0.99
    154. World's Fair0.98
    155. togetherai0.98
    156. Worlr0.93
    157. World's Fair0.97
    158. Unblocked1.00
    159. World's Fair0.98
    160. World's Fair0.96
    161. DATA0.98
    162. World's Fair0.99
    163. Resolve.ai1.00
    164. World's Far0.98
    165. World's Fa0.98
    166. WorldsFai0.84
    167. LanceDB1.00
    168. World's Fair0.96
    169. builderio0.95
    170. World'0.87
    171. Ravenna0.99
    172. World's Fair0.96
    173. CopilotKit1.00
    174. Wor1.00
    175. World's Fair0.96
    176. World's Fair0.99
    177. TOPK1.00
    178. World's Fair0.96
    179. co0.83
    180. World's Fair0.98
    181. eNCORD0.85
    182. World's Fair0.97
    183. rid'sFa0.96
    184. World's Fa0.95
    185. Microsoft0.92
    186. e0.71

  • 1:01 #6 done15 line(s)

    shot 6·sharpness 649.4

    1. GI Lab0.97
    2. World's Fai0.99
    3. Microsoft1.00
    4. AlEngineer0.95
    5. OpenAr0.92
    6. Fair1.00
    7. World's Fair0.97
    8. Engin1.00
    9. nerai0.96
    10. Wo."c0.88
    11. Akamai1.00
    12. AlEngineer0.96
    13. Fair1.00
    14. DATAC0.94
    15. World's Fair0.96

  • 1:07 #7 done10 line(s)

    shot 7·sharpness 636.1

    1. 10.62
    2. World's Fair0.96
    3. Google DeepMind0.95
    4. AlEngineer0.98
    5. r0.99
    6. ANTHRC0.98
    7. World's Fair0.98
    8. greptile1.00
    9. AlEngineer0.97
    10. World's Fair0.98

  • 1:17 #8 done15 line(s)

    shot 8·sharpness 626.9

    1. Lab1.00
    2. World's Fair0.99
    3. Microsoft1.00
    4. AlEngineer0.97
    5. OpenA'0.98
    6. ir1.00
    7. World's Fair0.97
    8. AIEn0.95
    9. ai1.00
    10. Worl1.00
    11. Akamai1.00
    12. AlEngineer0.96
    13. DATA1.00
    14. ir1.00
    15. World's Fair0.99

  • 1:40 #9 done16 line(s)

    shot 9·sharpness 2156.4

    1. AlEngineer0.97
    2. World'sFair1.00
    3. PRESENTED BY0.99
    4. Microsoft1.00
    5. Intrinsic1.00
    6. ilLab0.95
    7. World'sFair0.97
    8. air1.00
    9. oenAl0.95
    10. erai1.00
    11. Fair1.00
    12. On Al and Knowledge0.99
    13. air1.00
    14. TADOG1.00
    15. Pablo Castro / CVP & Distinguished Engineer0.98
    16. Microsoft1.00

  • 2:15 #10 done68 line(s)

    shot 10·sharpness 1203.1

    1. AlEngineer0.98
    2. World's Fair0.99
    3. ~19991.00
    4. ~20241.00
    5. © SPRENG.CPP ×0.93
    6. 2551.00
    7. ESULT CSpriteCngine::DOlenderALI( IDirectOradurface 'ps, RECT Ssource, RECT &target )0.89
    8. // Draw background0.94
    9. while( true ) (0.82
    10. MRESULT hr;0.90
    11. hrups->8ItFast( target.left, target.top, #_pbs, &source, DOBLTFAST_MOCOLORKEY );0.90
    12. vold Readide(vold" heile, int length, Snt count, uint32_t offsets, ist timeout, stdifunctioncvold(in0.85
    13. F (C > _OUTSTADIG_10)0.57
    14. DORD read_size - g_unbuffered ? page_align_size(length) : length;0.92
    15. throu std:rustime_error("foo many requests")10.80
    16. 1f( hr1-00_0K ) break;0.81
    17. for (int I = 0; I < count; $++) {0.83
    18. OMOR0 read_offset - g_unbuffered 7 page_align_offset (offsets[1]) : offsets[1];0.91
    19. PRESENTED BY1.00
    20. SF( hr1-ODERR_MASSTILLORAWENG) return he↓0.85
    21. entries 1 .overlapped.Internai a0.55
    22. 8_entries 1].overlapped.offset - read_offset;0.91
    23. 8_entries[1].overlapped.OffsetHägh = 0;0.89
    24. Microsoft0.99
    25. 2760.81
    26. STRSPR_MAPI:Sterator it;0.83
    27. for(tSprtDega(): StSprt() St)0.50
    28. cSprite *p=pr.second;0.89
    29. std::palr<std::string,(sprite*> pr-*it;0.67
    30. 18( p p->pSprite)0.83
    31. long lu, Sh, Sr, ia;0.79
    32. p->pSprite->get_Imagoseth( &iw):0.62
    33. _entries[1].completed - falso;0.90
    34. DD to_read - offsets[1] - read_offset + length > read_size 2 * read_size : read_size;0.93
    35. // Over-reading, tweak this to read only as many extra pages as0.97
    36. (IReadFile(hfile, R_entries[1].data, to_read, ML, &g_entries[1).overlapped))(0.91
    37. If (err I= ERROR_TO_PENDING)(0.85
    38. thros std::runtise error(ralled to read file");0.83
    39. p-spSprite->gut_Imagelumber( &is );0.68
    40. ssr.top-ir;0.97
    41. R_entries[1].completed = true;0.94
    42. int pending = coust;0.95
    43. AGI Lab1.00
    44. World's Fai0.99
    45. 1f( hrODER_SURFACELOST )(0.77
    46. hr=0ORestoreAl1();0.81
    47. 1f( hr1=00_oK) break;0.81
    48. for (int i = e; 1 < count; 1++) (0.87
    49. (_entries[1],completed)0.93
    50. uinta t* data a Rentrles[1] data;0.68
    51. 1f (g_unbuffered) (0.89
    52. callback(§, data);0.92
    53. pending…;0.88
    54. data offsets[1] - 8_entries[1].overlapped.offset;0.94
    55. 1f( hr!-0DERR_MASSTILLDRANING ) return hr;0.87
    56. ②0 0 live Sthee0.58
    57. In 254 Cal I Ta Siae4 UTF-E CRLF () C- 00.74
    58. ②00 lie Shere0.66
    59. (n 70, Cal1 To Soe 4 UTF-8 CRF () Ce 00.72
    60. sFair1.00
    61. leer0.84
    62. OpenAI0.92
    63. Engineer1.00
    64. therc0.92
    65. d'sFai1.00
    66. ieer0.89
    67. sFair1.00
    68. DATADOC0.96

  • 2:40 #11 done28 line(s)

    shot 11·sharpness 1902.4

    1. AlEngineer0.99
    2. World'sFair1.00
    3. OpenClaw-sized apps, not a single line written by hand0.99
    4. Opus 4.5 ships, agents write entire simple apps0.99
    5. Cursor launches. GitHub Copilot X expands to chat, PRs, etc.1.00
    6. IntelliSense: type-driven autocomplete1.00
    7. GitHub Copilot suggests whole lines & functions1.00
    8. PRESENTED BY0.96
    9. IntelliCode ranks completions with ML0.99
    10. Microsoft1.00
    11. 19961.00
    12. 20181.00
    13. 20211.00
    14. 20231.00
    15. 20251.00
    16. 20261.00
    17. AGI Lab1.00
    18. World'sFa1.00
    19. neer1.00
    20. sFair1.00
    21. OpenA1.00
    22. 'Engineer0.98
    23. ether0.99
    24. d'sFa0.99
    25. Engineering the future of Al0.98
    26. neer1.00
    27. sFair1.00
    28. DATADC1.00

  • 2:57 #12 done28 line(s)

    shot 12·sharpness 1930.9

    1. AlEngineer0.99
    2. World'sFair1.00
    3. OpenClaw-sized apps, not a single line written by hand1.00
    4. Opus 4.5 ships, agents write entire simple apps1.00
    5. Cursor launches. GitHub Copilot X expands to chat, PRs, etc.0.99
    6. IntelliSense: type-driven autocomplete1.00
    7. GitHub Copilot suggests whole lines & functions1.00
    8. PRESENTED BY0.97
    9. IntelliCode ranks completions with ML0.98
    10. Microsoft1.00
    11. 19961.00
    12. 20181.00
    13. 20211.00
    14. 20231.00
    15. 20251.00
    16. 20261.00
    17. 1AGI Lab0.95
    18. World'sF0.99
    19. ngineer1.00
    20. d's Fair0.97
    21. Open/0.94
    22. AlEngineer0.98
    23. geth0.94
    24. Norld'sF0.96
    25. Engineering the future of Al1.00
    26. ngineer1.00
    27. d'sFai1.00
    28. DATAD1.00

  • 3:35 #13 skipped

    shot 13·duplicate of #12

  • 3:44 #14 done24 line(s)

    shot 14·sharpness 1845.1

    1. AlEngineer1.00
    2. World's Fair0.99
    3. OpenClaw-sized apps, not a single line written by hand0.99
    4. Opus 4.5 ships, agents write entire simple apps0.99
    5. Cursor launches. GitHub Copilot X expands to chat, PRs, etc.1.00
    6. IntelliSense: type-driven autocomplete1.00
    7. GitHub Copilot suggests whole lines & functions1.00
    8. PRESENTED BY0.97
    9. IntelliCode ranks completions with ML0.99
    10. Microsoft1.00
    11. 19961.00
    12. 20181.00
    13. 20211.00
    14. 20231.00
    15. 20251.00
    16. 20261.00
    17. World's Fair0.97
    18. OpenA1.00
    19. We0.77
    20. AI0.78
    21. Wor0.82
    22. Engineering the future of Al1.00
    23. DAT1.00
    24. We0.79

  • 4:32 #15 done19 line(s)

    shot 15·sharpness 2134.2

    1. AlEngineer0.98
    2. World's Fair0.98
    3. General Availability0.99
    4. PRESENTED BY1.00
    5. Microsoft1.00
    6. Claude in Microsoft Foundry0.99
    7. Frontier models for reasoning, coding, and agentic workflows.1.00
    8. Claude Opus 4.80.97
    9. Claude Haiku 4.50.99
    10. il Lab0.93
    11. World's Fair0.99
    12. :air0.94
    13. OpenAl0.93
    14. gineer1.00
    15. erai1.00
    16. 's Fair0.89
    17. Engineering the future of Al1.00
    18. air1.00
    19. DATADOG0.99

  • 4:46 #16 done14 line(s)

    shot 16·sharpness 1579.4

    1. AlEngineer0.98
    2. World'sFair1.00
    3. PRESENTED BY0.97
    4. Microsoft1.00
    5. Extrinsic1.00
    6. World'sFair1.00
    7. OpenAl0.94
    8. Wc0.84
    9. AlEnginee0.99
    10. World's1.00
    11. Engineering the future of Al0.98
    12. 福0.53
    13. DATAD1.00
    14. Wc0.82

  • 5:38 #17 done25 line(s)

    shot 17·sharpness 2166.6

    1. AlEngineer1.00
    2. World's Fair0.97
    3. Your people. Your agents. Your IQ.0.98
    4. Interactive1.00
    5. Autonomous1.00
    6. People1.00
    7. Agents1.00
    8. PRESENTED BY1.00
    9. Microsoft1.00
    10. Microsoft IQ1.00
    11. Teams, roles,0.98
    12. State and actions1.00
    13. Curated1.00
    14. and workflows1.00
    15. of the business1.00
    16. knowledge1.00
    17. MAX1.00
    18. World'sFa1.00
    19. Fair1.00
    20. OpenAl0.93
    21. Engineer1.00
    22. d'sFa1.00
    23. Engineering the future of Al0.98
    24. Fair1.00
    25. Buildk1.00

  • 5:48 #18 done25 line(s)

    shot 18·sharpness 2176.5

    1. AlEngineer0.99
    2. World'sFair1.00
    3. Your people. Your agents. Your IQ.0.98
    4. Interactive1.00
    5. Autonomous1.00
    6. People1.00
    7. Agents1.00
    8. PRÉSENTED BY0.98
    9. Microsoft1.00
    10. Microsoft IQ1.00
    11. Teams, roles,1.00
    12. State and actions1.00
    13. Curated1.00
    14. and workflows1.00
    15. of the business1.00
    16. knowledge1.00
    17. Vorld's Fain0.96
    18. Amazc0.94
    19. OpenA1.00
    20. Wor1.00
    21. — AlEn0.79
    22. Vorl0.93
    23. Engineering the future of Al0.98
    24. Buil0.98
    25. Wor1.00

  • 6:19 #19 done37 line(s)

    shot 19·sharpness 3435.1

    1. AlEngineer0.98
    2. World's Fair0.98
    3. Microsoft IQ Platform1.00
    4. Unified intelligence for enterprise Al0.99
    5. PRESENTED BY0.99
    6. Microsoft1.00
    7. Work IQ0.99
    8. Fabric IQ1.00
    9. Foundry IQ1.00
    10. Web IQ1.00
    11. How your0.99
    12. How your0.97
    13. How your agents1.00
    14. How you connect to1.00
    15. employees work1.00
    16. business operates1.00
    17. unlock knowledge1.00
    18. web intelligence0.98
    19. Context on people,0.98
    20. Context on business1.00
    21. Context on policies,1.00
    22. Context from the web,1.00
    23. World's Fair0.94
    24. collaboration, and1.00
    25. entities, systems of1.00
    26. authoritative documents,1.00
    27. news, images and video1.00
    28. workflows1.00
    29. record, and actions1.00
    30. and knowledge bases1.00
    31. OpenAl0.89
    32. aka.ms/MicrosoftlQ0.99
    33. air1.00
    34. On Al and Knowledge0.99
    35. uildkite0.99
    36. Pablo Castro / CVP & Distinguished Engineer0.98
    37. Microsoft1.00

  • 7:07 #20 done40 line(s)

    shot 20·sharpness 2654.4

    1. AlEngineer0.99
    2. World's Fair0.98
    3. Microsoft IQ Platform1.00
    4. Unified intelligence for enterprise Al0.99
    5. PRESENTED BY1.00
    6. Microsoft1.00
    7. Work IQ0.99
    8. Fabric IQ1.00
    9. Foundry IQ1.00
    10. Web IQ1.00
    11. How your0.99
    12. How your1.00
    13. How your agents1.00
    14. How you connect to1.00
    15. employees work0.99
    16. business operates1.00
    17. unlock knowledge1.00
    18. web intelligence1.00
    19. Context on people,0.98
    20. Context on business1.00
    21. Context on policies,1.00
    22. Context from the web,1.00
    23. NIMAX1.00
    24. World'sF1.00
    25. collaboration, and0.99
    26. workflows1.00
    27. entities, systems of1.00
    28. record, and actions1.00
    29. authoritative documents,1.00
    30. and knowledge bases0.99
    31. news, images and video1.00
    32. 's Fair0.94
    33. neer1.00
    34. Open/0.89
    35. aka.ms/MicrosoftlQ0.99
    36. ineer-1.00
    37. )e0.63
    38. 's Fair0.95
    39. neer1.00
    40. 3uild0.95

  • 7:31 #21 done12 line(s)

    shot 21·sharpness 965.0

    1. AlEngineer0.99
    2. World's Fair0.99
    3. PRESENTED BY1.00
    4. A·B1.00
    5. Microsoft1.00
    6. cos(θ) =0.99
    7. |A||B|0.99
    8. IIMAX1.00
    9. World's Fa0.92
    10. Fair1.00
    11. ber0.98
    12. OpenA1.00

  • 8:00 #22 done

    shot 22·sharpness 1385.0

  • 8:20 #23 done

    shot 23·sharpness 2094.8

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

133 cues· 2,883 words· 15,612 chars

  1. 0:12 Now taking the stage is CVP and Distinguished Engineer at Microsoft, Pablo Castro.
  2. 0:22 Hello, everyone.
  3. 0:25 Hello, everyone.
  4. 0:26 Good morning.
  5. 0:28 It's great to be back here at the AI Engineer World's Fair.
  6. 0:33 Now, my job at Microsoft is to connect the dots between AI and knowledge.
  7. 0:40 As an information retrieval nerd, that's great for me.
  8. 0:43 I spend a lot of time looking at knowledge representation, extraction, search, and whatnot.
  9. 0:50 And thinking about agents and knowledge really invites to reflect on what it means to know something and the nature of how do we get things done based on what we know.
  10. 1:02 Next slide.
  11. 1:09 All right.
  12. 1:15 So this morning, what I thought we would do is spend a little bit of time talking about the nature of knowledge and split it into these three categories of intrinsic, extrinsic, and learned.
  13. 1:27 Intrinsic knowledge is just the knowledge that comes with the models.
  14. 1:30 It's what we train the models on, the training data, and what's stored in the models, kind of parametric memory.
  15. 1:38 And while it's kind of the obvious thing, I would argue this is the knowledge that actually threw us into the exponential we are in today.
  16. 1:45 It's what started many of the scenarios that then grew on all the things we're doing with agents today.
  17. 1:52 Let me give you an example with code.
  18. 1:54 So I wrote these two pieces of code about 25 years apart.
  19. 1:59 And yet, the process to put this thing together was surprisingly similar.
  20. 2:04 Like I had to sit down with what I knew or I had to go look up and then just write it up.
  21. 2:11 And while I'm illustrating this with knowledge, you could say the same thing about writing an email or creating a summary of a document.
  22. 2:19 Now, you can see this exponential at play in tasks like this, where I'm sure you can go further back, but an interesting point in time to start looking at this would be when Microsoft introduced IntelliSense.
  23. 2:30 It was in 96, and it was great.
  24. 2:33 You didn't have to remember function signatures anymore and whatnot.
  25. 2:36 It takes 22 years from there to go for the next step where machine learning helps us actually rank the options we give you in IntelliSense, so it's quicker to pick the right choice.
  26. 2:47 Just three years after that, GitHub Copilot launches.
  27. 2:51 And that was one key inflection point.
  28. 2:54 This was even before ChaiGPT was announced.
  29. 2:57 And I would argue that GitHub Copilot, ChaiGPT, that sort of experiences were heavily grounded on this intrinsic memory, what the models already knew.
  30. 3:06 From there, of course, things shifted, you know, a couple of years later, Courser launches, GitHub Copilot X launches, and how we do things kind of evolved really quick, which takes us to kind of late last year, Opus 4.5 ships, and then in rapid succession, you know, GPT, Opus, and other models keep getting better and better at coding, which takes us to early this year, where incredibly successful software like OpenClaw comes out to existence with not a single line of code written by hand.
  31. 3:35 So this is the shape of the exponential wearing, and a lot of this was powered by the intrinsic knowledge in models, and of course, their ability to reason.
  32. 3:47 Now, in the context of Microsoft, we want to make available all these models and make it easy for you to integrate them into the agents you're building.
  33. 3:54 We do this from our agent platform that starts in GitHub, where we all go and build.
  34. 4:00 It has a contextualization system so you can ground your agents.
  35. 4:04 And when it comes to agent hosting, observability, and management, we do all of these in Foundry.
  36. 4:10 Microsoft Foundry is also where we offer thousands of models in our model catalog, so you can pick whatever is the right model for the right task, and we keep adding more every day.
  37. 4:21 In fact, just yesterday we announced that Cloud in Microsoft Foundry is generally available, so you can use all the capabilities of Cloud in the context of the unified experience in Foundry, so you get the best of both worlds.
  38. 4:38 Now, intrinsic model got us here, but it only gets you so far if you're building a system or an agent that needs to participate in what's happening in an organization or a company.
  39. 4:49 And as an industry, we realized this early, and we saw the rag pattern emerge.
  40. 4:56 That started as a pretty low tech technique, but quickly evolved and what we do today with context engineering and it became a pretty sophisticated system for connecting agents and the knowledge they need to get their job done.
  41. 5:12 Of the many dimensions of which this got kind of complicated, I'm gonna pick on two.
  42. 5:18 One is kind of the evolution from simple and isolated data sets to whole company-wide grounding.
  43. 5:25 And the other one is how we started with simple vector search and whatnot, and we really saw this evolve into fairly complicated retrieval systems.
  44. 5:34 So let's start with company grounding.
  45. 5:36 Like at Microsoft, spending time with customers, one of the things we saw early was that whenever you build an agent, you always have the knowledge you care about for that agent and you manage that yourself, but you also need to ground the agent often on the kind of ambient data of your organization whenever the agent leaves.
  46. 5:54 This includes maybe your documents, your emails, your chat threads, or the information in your data warehouse and whatnot.
  47. 6:03 So we built Microsoft IQ as a way to give you a single entry point into all these kind of ambient data that agents need to get their job done in addition to the specific information that you build into the agent.
  48. 6:16 Microsoft IQ is not one feature, it's more like a set of capabilities that goes from work IQ that connects your agents to all the documents in, say, SharePoint, all the emails, calendar, your chats, and the connections between people, to Fabric IQ that gives you access to all your analytics assets, from data warehouses and data lakes to Power BI reports, and Foundry IQ, which is what you use for your org agents, where you can push your own data and then use it for grounding.
  49. 6:43 And of course, sometimes you have your agents need to go out to the web to ground on data, maybe not yours, it's public information, but you need to use it to complete the picture of what the agent world view is, and for that we have WebIQ.
  50. 7:01 Now, this first part allows agents to ground on kind of this ambient data.

Chapters

  1. 0:00 Introduction and speaker background
  2. 1:14 Defining the nature of knowledge: Intrinsic, Extrinsic, and Learned
  3. 1:27 Intrinsic knowledge and the history of AI coding tools
  4. 4:38 Extrinsic knowledge and corporate data grounding
  5. 7:06 Evolution of retrieval systems and Foundry IQ
  6. 9:56 Foundry IQ demo: Building a knowledge base
  7. 13:08 Learned knowledge: The agent learning loop
  8. 14:25 Foundry agent optimization demo
  9. 16:49 Closing remarks and resources
  10. 17:03 "This is a real learning loop materialized in practice... we can enable this learning loops that will capture this differentiated capability that lives in each one of the companies and organizations we work on." (16:40

Open at this second