Это пробный урок. Оформите подписку, чтобы получить доступ ко всем материалам курса. Премиум

  1. Урок 1. 00:11:47
    Who Is This Course For
  2. Урок 2. 00:12:16
    The Autonomy Spectrum Direct Call, Workflow, Single Agent, Multi Agent
  3. Урок 3. 00:11:38
    The Agent Loop
  4. Урок 4. 00:12:54
    Anatomy of an Agent System
  5. Урок 5. 00:13:24
    Compounding Error The Arithmetic of Reliability
  6. Урок 6. 00:13:31
    Workflows vs Agents The First Design Decision
  7. Урок 7. 00:13:30
    The Design Worksheet
  8. Урок 8. 00:11:32
    The Pattern Map
  9. Урок 9. 00:12:35
    Prompt Chaining
  10. Урок 10. 00:14:02
    Routing
  11. Урок 11. 00:12:24
    Parallelization Sectioning and Voting
  12. Урок 12. 00:13:05
    Evaluator Optimizer
  13. Урок 13. 00:14:10
    Structured Output as a Contract
  14. Урок 14. 00:13:25
    Composing Workflow Patterns
  15. Урок 15. 00:13:02
    Function Calling The Request, Execute, Return Cycle
  16. Урок 16. 00:14:20
    Designing a Tool the Model Can Use
  17. Урок 17. 00:14:48
    Tool Granularity, Namespacing, and Dynamic Selection
  18. Урок 18. 00:14:00
    Tool Results What a Response Should Contain and What an Error Should Say
  19. Урок 19. 00:13:42
    Code Execution as a General Tool
  20. Урок 20. 00:12:54
    Retrieval as a Tool Searching on Demand and Agentic RAG
  21. Урок 21. 00:13:27
    Acting on the Outside World Read and Write Tools and Reversibility
  22. Урок 22. 00:13:56
    Case Study The Tool Set for Desk
  23. Урок 23. 00:13:36
    Reasoning Between Actions Prompted and Native
  24. Урок 24. 00:13:05
    Task Decomposition and Explicit Task Lists
  25. Урок 25. 00:13:15
    Plan Then Execute
  26. Урок 26. 00:12:50
    Replanning Detecting Drift and Stalled Progress
  27. Урок 27. 00:14:13
    Reflection and Self Critique
  28. Урок 28. 00:13:31
    Exploration Multiple Candidates, Scoring, and Pruning
  29. Урок 29. 00:12:25
    Choosing a Planning Pattern
  30. Урок 30. 00:13:29
    The Context Window as a Budget
  31. Урок 31. 00:12:18
    How Specific a System Prompt Should Be
  32. Урок 32. 00:14:11
    Compaction Summarizing History Without Losing the Task
  33. Урок 33. 00:12:37
    Context Resets and Structured Handoffs
  34. Урок 34. 00:12:11
    Instruction Files and Scoped Context Assembly
  35. Урок 35. 00:12:32
    Deterministic Lifecycle Hooks
  36. Урок 36. 00:13:16
    Skills Packaged Procedures with Progressive Disclosure
  37. Урок 37. 00:12:14
    Autonomy Levels per Action
  38. Урок 38. 00:12:12
    Approval Checkpoints and Risk Classification
  39. Урок 39. 00:13:38
    Approval Fatigue When Reviewers Stop Reading
  40. Урок 40. 00:12:38
    Escalation The Agent Hands Off When Uncertain
  41. Урок 41. 00:12:37
    Steering, Interruption, and Pause Resume
  42. Урок 42. 00:13:39
    Feedback Capture Corrections Become Evals and Memory
  43. Урок 43. 00:13:06
    Case Study Checkpoints for Desk
  44. Урок 44. 00:12:32
    How Agents Fail A Taxonomy
  45. Урок 45. 00:13:39
    Guardrail Layering Input, Tool Call, Output
  46. Урок 46. 00:13:21
    Outcome Verification Check the State, Not the Claim
  47. Урок 47. 00:14:36
    Grounding Citations and Checking Claims Against Sources
  48. Урок 48. 00:12:12
    Least Privilege Permission Design and Scoped Credentials
  49. Урок 49. 00:13:03
    Identity and Delegation Acting on Behalf of a User
  50. Урок 50. 00:14:09
    Sandboxing and Containment
  51. Урок 51. 00:12:15
    Execution Budgets Steps, Time, Tokens, Money
  52. Урок 52. 00:12:45
    Circuit Breakers and Stop Controls
  53. Урок 53. 00:13:19
    Durable State Checkpoints, Resume, and Side Effects
  54. Урок 54. 00:13:02
    Exception Handling and Recovery
  55. Урок 55. 00:13:32
    Prompt Injection The Core Problem
  56. Урок 56. 00:13:42
    Where Injection Enters Content, Tool Descriptions, Memory, Other Agents, Skills
  57. Урок 57. 00:13:42
    Three Properties That Make an Agent Exploitable The Lethal Trifecta and the Rule of Two
  58. Урок 58. 00:13:38
    Dual LLM Privileged Planner, Quarantined Reader
  59. Урок 59. 00:12:48
    Restrict What Can Act Action Selector, Plan Then Execute, Context Minimization
  60. Урок 60. 00:13:17
    Isolate What Reads Map Reduce and Code Then Execute
  61. Урок 61. 00:14:44
    Probabilistic Defenses What the Model Gives You and What You Add
  62. Урок 62. 00:13:58
    Other Agent Risks Supply Chain, Code Execution, and Misplaced Human Trust
  63. Урок 63. 00:13:49
    Case Study Securing Mailbox With Web Access
  64. Урок 64. 00:14:25
    Why Agent Evaluation Is Different, and Its Vocabulary
  65. Урок 65. 00:13:02
    The First Eval Set From Real Failures
  66. Урок 66. 00:13:46
    Grader Types Code Based, Model Based, Human
  67. Урок 67. 00:13:49
    Evaluating Trajectories
  68. Урок 68. 00:12:48
    Capability Evals vs Regression Evals
  69. Урок 69. 00:13:52
    Tracing and Trajectory Logging
  70. Урок 70. 00:12:09
    Debugging a Failed Run From Its Trace
  71. Урок 71. 00:14:25
    Eval Driven Development
  72. Урок 72. 00:13:30
    Case Study The Eval Set for Desk
  73. Урок 73. 00:13:12
    Kinds of Memory, and What to Store, Recall, and Forget
  74. Урок 74. 00:12:54
    Tiered Memory and Just in Time Loading
  75. Урок 75. 00:12:46
    Structured Note Taking Scratchpads, Task Files, Progress Files
  76. Урок 76. 00:13:46
    Retrieval Backed Memory Files vs Embedding Stores
  77. Урок 77. 00:13:25
    Memory Consolidation and Staleness
  78. Урок 78. 00:12:00
    Case Study Forge Across Many Sessions
  79. Урок 79. 00:14:01
    Case Study Mailbox for One User Over Months
  80. Урок 80. 00:13:47
    When One Agent Is Not Enough
  81. Урок 81. 00:14:12
    Orchestrator Workers
  82. Урок 82. 00:14:26
    Context Isolated Sub Agents and Fork Join
  83. Урок 83. 00:12:34
    Sequential Pipelines and Handoffs
  84. Урок 84. 00:13:17
    Group Chat, Debate, and Maker Checker Loops
  85. Урок 85. 00:13:49
    Communication Contracts Shared State vs Message Passing
  86. Урок 86. 00:12:54
    Failure Containment and the Cost of Coordination
  87. Урок 87. 00:12:57
    Case Study Scout as Orchestrator Workers
  88. Урок 88. 00:12:49
    The Token Economics of an Agent Loop
  89. Урок 89. 00:12:37
    Prompt Caching and Cache Friendly Loop Design
  90. Урок 90. 00:13:14
    Model Routing and Tiering
  91. Урок 91. 00:12:38
    Tool Result Caching and Deduplication
  92. Урок 92. 00:13:01
    Latency Patterns Streaming, Parallel Tool Calls, Speculative Work
  93. Урок 93. 00:12:55
    Serving Infrastructure Queues, Rate Limits, and Rainbow Deployments
  94. Урок 94. 00:13:27
    Background Agents Triggers, Isolation, and Reporting Back
  95. Урок 95. 00:12:47
    Case Study Desk's Cost Reduced
  96. Урок 96. 00:12:30
    Model Context Protocol Servers, Clients, Tools, Resources, Prompts
  97. Урок 97. 00:12:38
    Agent to Agent Protocols Agent Cards, Delegation, Task Streaming
  98. Урок 98. 00:14:16
    What a Framework Gives You Build or Adopt
  99. Урок 99. 00:12:50
    Designing for Portability Keeping the Pattern Independent of the Vendor
  100. Урок 100. 00:13:01
    Retiring Harness Parts When Models Improve
  101. Урок 101. 00:14:11
    The Completed Worksheet
  102. Урок 102. 00:13:45
    Design a Customer Support Agent
  103. Урок 103. 00:12:46
    Design a Coding Agent
  104. Урок 104. 00:13:56
    Design a Data Analysis Agent
  105. Урок 105. 00:12:43
    Design a Browser Automation Agent
  106. Урок 106. 00:13:30
    Design an Incident Response Agent
  107. Урок 107. 00:14:05
    Capstone A Blank Brief