📄️ AI agents
How an agent decides, calls tools, remembers and stays within a budget. 6 hours, a research-agent build project, a 40-question exam and a verifiable certificate.
📄️ 1. What an agent adds
Module 1 of the AI Agents premium course: the honest difference between a call, a chain and an agent, the real price of autonomy in tokens and latency, and when an agent is the wrong answer.
📄️ 2. Reasoning and acting loop
Module 2 of the AI Agents premium course: the ReAct pattern explained step by step, the stopping condition that avoids infinite loops, and a runnable Python implementation in sixty lines.
📄️ 3. Function calling and tools
Module 3 of the AI Agents premium course: the JSON schema of a tool, why the description does most of the work, how many tools an agent can juggle, and validating arguments before execution.
📄️ 4. Memory
Module 4 of the AI Agents premium course: how the loop's context saturates, when to summarise, storing durable memories with embeddings, and what an agent must be told to forget.
📄️ 5. Planning
Module 5 of the AI Agents premium course: writing an explicit plan before acting, replanning after new evidence, sub-agents by role, and knowing when planning is worth its cost.
📄️ 6. Self-critique
Module 6 of the AI Agents premium course: reflection prompts, a second model as verifier, tool-based verification through recalculation, and the honest limits of self-correction.
📄️ 7. Guardrails
Module 7 of the AI Agents premium course: iteration and spend caps, per-tool permissions, human confirmation before irreversible actions, and the sandbox that isolates untrusted content.
📄️ 8. Typical failures
Module 8 of the AI Agents premium course: a catalogue of the failures observed on the running agent, with the trace, the root cause and the fix for each, including prompt injection through fetched pages.
📄️ 9. Observability
Module 9 of the AI Agents premium course: the full trace of an agent run, per-step cost accounting, replay from a stored trace, and a side-by-side comparison with the same agent rewritten in LangGraph.
📄️ 10. Project
Module 10 of the AI Agents premium course: assembling every module into a research agent, running a 30-question evaluation, measuring accuracy and cost, and preparing the production checklist.
📄️ Recap and exam
Complete recap of the AI Agents premium course: reasoning loop, tools, memory, planning, verification, guardrails, failures and observability, then the 40-question exam.