vidtheque.

read-only demo

Videos Z-c11pV_uvU

Anthropic's CCA Exam as a Field-Guide for Agentic Engineering — Frank Coyle, UC Berkeley

index_state ready data_status ok

AI Engineer· published 2026-08-08· 0:20:08· en-US· indexed 2026-08-10 19:34

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 0:29, 1 of 1 keyframes kept
  5. Shot 4, 0:29 to 1:00, 1 of 1 keyframes kept
  6. Shot 5, 1:00 to 1:31, 0 of 1 keyframes kept
  7. Shot 6, 1:31 to 1:35, 1 of 1 keyframes kept
  8. Shot 7, 1:35 to 1:59, 1 of 1 keyframes kept
  9. Shot 8, 1:59 to 2:43, 0 of 1 keyframes kept
  10. Shot 9, 2:43 to 3:23, 1 of 1 keyframes kept
  11. Shot 10, 3:23 to 4:11, 1 of 1 keyframes kept
  12. Shot 11, 4:11 to 4:40, 1 of 1 keyframes kept
  13. Shot 12, 4:40 to 5:08, 0 of 1 keyframes kept
  14. Shot 13, 5:08 to 5:37, 0 of 1 keyframes kept
  15. Shot 14, 5:37 to 6:05, 0 of 1 keyframes kept
  16. Shot 15, 6:05 to 6:31, 1 of 1 keyframes kept
  17. Shot 16, 6:31 to 6:56, 0 of 1 keyframes kept
  18. Shot 17, 6:56 to 7:22, 0 of 1 keyframes kept
  19. Shot 18, 7:22 to 7:47, 0 of 1 keyframes kept
  20. Shot 19, 7:47 to 8:17, 0 of 1 keyframes kept
  21. Shot 20, 8:17 to 8:42, 0 of 1 keyframes kept
  22. Shot 21, 8:42 to 9:07, 0 of 1 keyframes kept
  23. Shot 22, 9:07 to 9:33, 0 of 1 keyframes kept
  24. Shot 23, 9:33 to 9:58, 0 of 1 keyframes kept
  25. Shot 24, 9:58 to 10:23, 0 of 1 keyframes kept
  26. Shot 25, 10:23 to 10:48, 0 of 1 keyframes kept
  27. Shot 26, 10:48 to 11:14, 0 of 1 keyframes kept
  28. Shot 27, 11:14 to 11:34, 1 of 1 keyframes kept
  29. Shot 28, 11:34 to 11:59, 0 of 1 keyframes kept
  30. Shot 29, 11:59 to 12:24, 0 of 1 keyframes kept
  31. Shot 30, 12:24 to 12:49, 0 of 1 keyframes kept
  32. Shot 31, 12:49 to 13:15, 0 of 1 keyframes kept
  33. Shot 32, 13:15 to 13:40, 0 of 1 keyframes kept
  34. Shot 33, 13:40 to 14:11, 0 of 1 keyframes kept
  35. Shot 34, 14:11 to 14:43, 0 of 1 keyframes kept
  36. Shot 35, 14:43 to 15:14, 0 of 1 keyframes kept
  37. Shot 36, 15:14 to 15:49, 0 of 1 keyframes kept
  38. Shot 37, 15:49 to 16:16, 0 of 1 keyframes kept
  39. Shot 38, 16:16 to 16:44, 0 of 1 keyframes kept
  40. Shot 39, 16:44 to 17:12, 0 of 1 keyframes kept
  41. Shot 40, 17:12 to 17:40, 0 of 1 keyframes kept
  42. Shot 41, 17:40 to 18:07, 0 of 1 keyframes kept
  43. Shot 42, 18:07 to 18:40, 0 of 1 keyframes kept
  44. Shot 43, 18:40 to 19:13, 1 of 1 keyframes kept
  45. Shot 44, 19:13 to 19:19, 0 of 1 keyframes kept
  46. Shot 45, 19:19 to 19:51, 1 of 1 keyframes kept
  47. Shot 46, 19:51 to 20:07, 0 of 1 keyframes kept

47 shot(s).

keyframes kept every frame deduplicated

What was stored

cues
232
whisperx 232
chunks
35
from 232 cues
keyframes
14
kept of 47 captured
frames with text
14
392 lines read
chapters
0
from the source metadata
keyframe bytes
6.9 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 — 2026-08-09 21:06 0s
stt done — 2026-08-08 23:29 25s
chunk done — 2026-08-08 23:29 0s
text_embed done — 2026-08-10 19:34 0s
keyframe done — 2026-08-08 23:29 2m 42s
ocr done — 2026-08-08 23:32 9s
frame_embed done — 2026-08-10 19:34 2s

Frames, and what the machine read

  • 0:02 #0 done2 line(s)

    shot 0·sharpness 451.2

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

  • 0:03 #1 done2 line(s)

    shot 1·sharpness 662.6

    1. AlEngineer0.96
    2. World's Fair0.99

  • 0:10 #2 done25 line(s)

    shot 2·sharpness 2686.6

    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. arize1.00
    9. aws1.00
    10. Braintrust1.00
    11. bright data0.99
    12. B1.00
    13. Browserbase1.00
    14. docker1.00
    15. :neo4j0.93
    16. ORACLE1.00
    17. PayPal1.00
    18. qodo1.00
    19. reducto1.00
    20. Sonar1.00
    21. Makers of0.99
    22. togetherai1.00
    23. Unblocked1.00
    24. WorkOS1.00
    25. SonarQube1.00

  • 0:14 #3 done2 line(s)

    shot 3·sharpness 243.1

    1. AlEngineer0.96
    2. World's Fair0.97

  • 0:50 #4 done27 line(s)

    shot 4·sharpness 2111.6

    1. World'sFair1.00
    2. AlEngineer0.97
    3. DEFAULT1.00
    4. Slide Layout0.97
    5. Appearance1.00
    6. Title0.99
    7. Body0.99
    8. Slide Number1.00
    9. Background1.00
    10. ANTHROPIC CERTIFICATION1.00
    11. · FIELD GUIDE0.95
    12. Standard1.00
    13. Dynamic0.97
    14. Current Fil0.95
    15. PRESENTED BY0.99
    16. The Claude Certified Architect (CCA)1.00
    17. Color Fll0.99
    18. Microsoft1.00
    19. Exam1.00
    20. Inside the CCA Exam: A Field Guide to Building Agentic Systems1.00
    21. Frank P. Coyle, PhD1.00
    22. UC Berkeley0.99
    23. Al Engineer World's Fair1.00
    24. July 3, 20260.98
    25. Edit Slide Layout0.93
    26. Engineering the future of Al0.99
    27. World'sFair1.00

  • 1:21 #5 skipped

    shot 5·duplicate of #4

  • 1:32 #6 done21 line(s)

    shot 6·sharpness 2091.3

    1. 00.61
    2. AlEngineer0.99
    3. World's Fair0.98
    4. Image Styles1.00
    5. Border1.00
    6. ANTHROPIC CERTIFICATION1.00
    7. FIELD GUIDE1.00
    8. The Claude Certified Architect (CCA)1.00
    9. Reflection0.95
    10. Opacity0.96
    11. Exam1.00
    12. Title0.97
    13. Inside the CCA Exam: A Field Guide to Building Agentic Systems1.00
    14. Caption1.00
    15. Frank P. Coyle, PhD1.00
    16. UC Berkeley1.00
    17. Al Engineer World's Fair1.00
    18. July 3, 20260.98
    19. TRACK 8· JULY 2, 20260.95
    20. World's Fair1.00
    21. Agentic Engineering1.00

  • 1:54 #7 done35 line(s)

    shot 7·sharpness 2338.0

    1. AlEngineer0.98
    2. Slide Layout0.97
    3. World'sFair1.00
    4. DEFAULT1.00
    5. MY TEACHING PHILOSOPHY1.00
    6. Appearance1.00
    7. Title1.00
    8. GG0.61
    9. Body0.98
    10. Slide Number1.00
    11. Background1.00
    12. Standard1.00
    13. Dynamic0.99
    14. Nothing is a mistake.0.99
    15. Current Fil0.96
    16. Color Fal0.94
    17. There's no win and no fail.1.00
    18. There's only make.1.00
    19. Sister Corita Kent · Rule 6, Some Rules for Students and Teachers (1967)0.98
    20. artist, educator, and nun — popularized by composer John Cage0.98
    21. Run the code, break it, leam from what0.97
    22. Experiment freely1.00
    23. happens.1.00
    24. Build to understand1.00
    25. We leam Al by making with it — not just0.97
    26. reading about it.1.00
    27. 81.00
    28. No final grade on a fist attempt; only the next1.00
    29. Iterate without fear0.99
    30. version.1.00
    31. Before we get into the exam, a word on how I teach – because it shepes how Il present llof this.0.97
    32. Edit Slide Layout0.94
    33. TRACK 8• JULY 2, 20260.95
    34. World's Fair0.98
    35. Agentic Engineering1.00

  • 2:21 #8 skipped

    shot 8·duplicate of #7

  • 2:52 #9 done30 line(s)

    shot 9·sharpness 2667.1

    1. World'sFair1.00
    2. AlEngineer0.98
    3. DEFAULT1.00
    4. Slide Layout0.98
    5. The Claude Certified Architect (CCA) Exam1.00
    6. Appearance0.99
    7. Title0.99
    8. Body0.97
    9. Slide Number0.99
    10. Background1.00
    11. Released March 12, 20260.98
    12. Standard1.00
    13. Dynamic0.99
    14. Anthropic's first official technical certification (Foundations / CCA-F), part of the Claude Partner Network.0.99
    15. Current Fill0.95
    16. Color FIl0.87
    17. 21.00
    18. Scenario-based and timed1.00
    19. 60 multiple-choice questions in 120 minutes, with a passing score of 720 on a scaled 100–1,000 range.0.99
    20. Judgment under realistic constraints1.00
    21. 31.00
    22. Every question drops you into a production situation — a broken loop, a poorly scoped tool, a system under load — and0.99
    23. asks for the best architectural call. You cannot memorize your way through it.0.99
    24. The takeaway: the exam is a curriculum in disguise — what it tests is what production agentic systems actually demand.0.98
    25. Quick orientation before we dive into the soenarios. It's a Curriculum0.96
    26. Released March 12, 2026 – Anthropic's fist technical oer, the Foundations level, gated trough the Partner Network at launch.0.96
    27. Edit Slide Layout0.99
    28. TRACK 8· JULY 2,20260.94
    29. World's Fair1.00
    30. Agentic Engineering1.00

  • 4:05 #10 done47 line(s)

    shot 10·sharpness 2770.0

    1. 00.50
    2. 153%0.99
    3. World's Fair0.96
    4. AlEngineer0.97
    5. DEFAULT1.00
    6. Slide Layout0.96
    7. The Five Exam Domains1.00
    8. Weighted by share of the exam1.00
    9. Appearance1.00
    10. Title0.99
    11. Body0.90
    12. Slide Number1.00
    13. DOMAIN 10.99
    14. Agentic Architecture & Orchestration1.00
    15. Loops and control flow; stop_reason — when to end vs. continue; guardrails and enforcement via hooks.0.99
    16. 27%1.00
    17. Background1.00
    18. Standard0.97
    19. Dynamic1.00
    20. Current Fll0.96
    21. Claude Code Configuration & Workflows1.00
    22. DOMAIN 31.00
    23. 20%1.00
    24. Color Fal0.90
    25. CLAUDE.md hierarchy; custom commands and skills; CI/CD and batch use; automation and repeatable0.97
    26. execution.1.00
    27. DOMAIN 41.00
    28. Prompt Engineering & Structured Output1.00
    29. 20%1.00
    30. Clear instructions, constraints, and success conditions; structured output via tool_use and JSON schemas.1.00
    31. DOMAIN 21.00
    32. Tool Design & MCP Integration0.98
    33. 18%1.00
    34. Make tools explicit, narrow, and easy to select correctly; structured errors; MCP server setup and0.99
    35. configuration.1.00
    36. DOMAIN 50.99
    37. Context Management & Reliability1.00
    38. 15%1.00
    39. refresh strategies.1.00
    40. Context positioning — what to place early vs. late; long-session reliability: summarization, degradation risks,0.98
    41. Read the weights: architecture and orchestration is the heaviest slice — agentic Al is a systems-design problem first.0.99
    42. The five domains, ordered by weight – and the weighting itself tels a story.0.96
    43. Domain 1, Agentic Architecture & Orchestration, is 27% — more than a quarter of the exam Loops, controlfow, stop_reason, guardrals vla hooks. The single heaviest area.0.89
    44. Edit Slide Layout0.99
    45. TRACK 8· JULY 2,20260.96
    46. World's Fair0.99
    47. Agentic Engineering1.00

  • 4:36 #11 done55 line(s)

    shot 11·sharpness 2631.9

    1. AlEngineer0.99
    2. Slide Layout0.98
    3. World'sFair1.00
    4. DEFAULT1.00
    5. The Six Production Scenarios1.00
    6. 4 of 6 drawn per exam0.98
    7. Appearance1.00
    8. Title0.95
    9. Body0.95
    10. Slide Number0.99
    11. SCENARIO 10.99
    12. Customer Support Resolution Agent1.00
    13. D1-50.94
    14. /0.78
    15. SCENARIO 21.00
    16. Code Generation with Claude Code0.98
    17. D2-30.99
    18. Background1.00
    19. Standard1.00
    20. Dynamic0.99
    21. Agentic loop control, stop_reson handling, and escalation1.00
    22. ClAUDE.md hierarchy, plan mode, and sooped project0.97
    23. Current Fil0.96
    24. to a human.1.00
    25. context.0.99
    26. Color Fall0.88
    27. SCENARIO 31.00
    28. D1·50.85
    29. SCENARIO 40.99
    30. D3-50.98
    31. Multi-Agent Research System1.00
    32. Developer Productivity with Claude1.00
    33. Hub-and-spoke orchestration, tool budgets, and context1.00
    34. Subtask isolation, context compaction, long-session0.98
    35. isolation.1.00
    36. reliability.1.00
    37. SCENARIO 51.00
    38. Claude Code for Cl/CD0.99
    39. D3·50.93
    40. 田1.00
    41. SCENARIO 61.00
    42. Structured Data Extraction0.98
    43. D4·50.95
    44. Headless -p execution, JSON output, batch vs. blocking1.00
    45. Schema enforcement via too_use, validation, and0.99
    46. latency.1.00
    47. confidence gates.1.00
    48. On exam day: 4 of these 6 are chosen at random and every question is anchored to them. (Anthropíc may expand the pool — verify the current Exam0.99
    49. Guide.)0.96
    50. These six production scenarios are the backbone of the exam – and the structure of the rest of my talk.0.99
    51. On any given sitting, 4 of the 6 are randomly selected, and EVERY question is anchored to one of those soenarios. You/re never answering in a vacuum - you're answering as the architect of a specific production system0.98
    52. Edit Slide Layout0.96
    53. TRACK 8· JULY 2,20260.96
    54. World's Fair0.99
    55. Agentic Engineering1.00

  • 5:05 #12 skipped

    shot 12·duplicate of #11

  • 5:31 #13 skipped

    shot 13·duplicate of #11

  • 5:40 #14 skipped

    shot 14·duplicate of #11

  • 6:08 #15 done34 line(s)

    shot 15·sharpness 2698.1

    1. World'sFair1.00
    2. AlEngineer0.99
    3. CS keeps rediscovering the loop0.99
    4. DEFAULT1.00
    5. Slide Layout0.98
    6. Appearance0.99
    7. Title0.94
    8. Body0.96
    9. Slide Number0.99
    10. BÖHM-JACOPINI (1966) — every computation needs three things0.99
    11. Background1.00
    12. Standard1.00
    13. Dynamic0.99
    14. PRESENTED BY1.00
    15. SEQUENCE1.00
    16. do this, then that0.98
    17. CONDITIONAL1.00
    18. if/then1.00
    19. LOOP1.00
    20. repeat until done1.00
    21. Current Fill0.94
    22. Color Fal0.91
    23. Microsoft1.00
    24. 60 years later ↓0.95
    25. THE NEW PARADIGM IS THE LOOP1.00
    26. "My job is to write loops."0.98
    27. "Design loops that prompt your agents."0.98
    28. Boris Cherny · creator of Claude Code, Anthropic0.99
    29. Peter Steinberger - creator of OpenClaw0.98
    30. sequece and condoals he p iswhatsew Ne sdeu thelops y here rks.0.54
    31. Edit Slide Layout0.96
    32. TRACK 8• JULY 2, 20260.96
    33. World's Fair0.98
    34. Agentic Engineering1.00

  • 6:41 #16 skipped

    shot 16·duplicate of #15

  • 6:59 #17 skipped

    shot 17·duplicate of #15

  • 7:40 #18 skipped

    shot 18·duplicate of #15

  • 8:05 #19 skipped

    shot 19·duplicate of #10

  • 8:34 #20 skipped

    shot 20·duplicate of #9

  • 9:04 #21 skipped

    shot 21·duplicate of #9

  • 9:15 #22 skipped

    shot 22·duplicate of #9

  • 9:40 #23 skipped

    shot 23·duplicate of #9

Transcript

232 cues· 2,753 words· 14,677 chars

  1. 0:12 Okay, we're getting rolling, and welcome aboard.
  2. 0:15 Just had a little technical issues, but we'll resolve them.
  3. 0:18 So my name is Frank Coyle.
  4. 0:21 I am a computer science guy.
  5. 0:23 I've been teaching computer science for over 30 years, and I'm now teaching at Berkeley.
  6. 0:28 And one of the problems that all my students,
  7. 0:32 past and present are having is AI, because computer science is no longer the magic pathway to a job.
  8. 0:40 So I've been trying to figure out ways to help them, come up with schemes to help them get ready for this world of agentic AI.
  9. 0:49 And one of the things that sort of dropped into my plate was something called the Claude Certified Architect Exam.
  10. 0:58 which I will be talking about today, and it has a number of aspects to it, and I think if you're interested in a career in agentic AI, to certainly take a look at least what the exam is about, because I feel that Anthropic knows how people are using their system and what the issues are gonna be.
  11. 1:19 So before we jump into that, I wanna give a little bit of my philosophy.
  12. 1:26 What, what, what, what?
  13. 1:33 May have to do this manually, getting stuck.
  14. 1:36 So, this is a quote from a woman named Sister Corita Kent.
  15. 1:42 Nothing is a mistake, there's no win and no fail, there's only make.
  16. 1:48 Bottom line here is experiment, experiment, experiment.
  17. 1:52 Not only should you read, but you should do, you should make stuff.
  18. 1:56 Now, what happens when you make stuff?
  19. 1:59 A lot of times, things don't work.
  20. 2:03 Thomas Edison said, I have not failed.
  21. 2:07 I've only found 10,000 ways that don't work.
  22. 2:13 What I want to emphasize here is that what this shows us are something that in the design patterns movement, which came around in the early 1990s with object-oriented programming, we had patterns for objects, we now have patterns for agents, but there's also anti-patterns, and I think anti-patterns are a key to understanding what you should not do, because understanding what you should not do is the key to leading you to what you should do.
  23. 2:44 So, a little bit about the CLAWD certified exam, released in March, so it's brand new.
  24. 2:50 It is based on scenarios, it is timed, it is proctored, it is available to companies in the CLAWD ecosystem, the anthropic ecosystem, but individuals can pay $99 and take the exam once every six months.
  25. 3:11 And it's not just,
  26. 3:13 multiple choice questions, it is multiple choice, but they are based on realistic constraints and realistic scenarios.
  27. 3:24 The five domains.
  28. 3:26 There are five domains that are covered and they give you the percentages of each.
  29. 3:30 So agentic architecture, 27%.
  30. 3:33 Cloud code, how to configure the cloud code system and workflow, 20%.
  31. 3:40 Doing prompt engineering, structuring your output using JSON all over the place.
  32. 3:47 tool design, model context protocol integration.
  33. 3:51 These are topics that you should understand and know whether you're gonna take the exam or not.
  34. 3:56 This is gonna help you get ready for whatever the agentic world is gonna throw at you.
  35. 4:02 And then there's gonna be context management and reliability.
  36. 4:06 So these are the areas of the kind of questions you're gonna run into.
  37. 4:13 Then there are, and they provide you with six production scenarios, and the exam will randomly choose four, and all the questions will be centered around the four that they choose.
  38. 4:29 And what I'm gonna do is walk you through
  39. 4:34 the production scenarios and give you some anti-patterns to be aware of.
  40. 4:38 Because there's a number of ways you can solve the problem, but one of the big things is what not to do.
  41. 4:43 And that often can be the key to getting these questions right.
  42. 4:47 So number one, customer support resolution agent.
  43. 4:50 So we have agentic loops.
  44. 4:52 Control, something called stop reason, which is what Claude Code has every time something happens, there's a stop reason and you need to take a look at that because that can give you a lot of information about what's going on.
  45. 5:05 Scenario two, code generation.
  46. 5:08 Three, multi-agent research system, which we'll look at.
  47. 5:12 How do you distribute your agents?
  48. 5:16 Hub and spoke, who's the orchestrator, how much information should they know?
  49. 5:20 All these are important factors.
  50. 5:23 Scenario four, developer,

Open at this second