Course
Full syllabus, tracks, graduation criteria, and a module-to-practice map.
What you get
Runnable Examples
Minimal Python examples for agents, tools, MCP, memory, workflow, colonies, evals, and RAG.
Benchmarks
Lightweight behavior checks for tool use, RAG, workflow, security, and observability.
Lesson Plans
Instructor-ready 90-minute teaching plans for every module.
Study Groups
4-week, 8-week, and workshop formats for teams and learning cohorts.
Showcases
Healthcare, finance, and enterprise demos with sample outputs.
Domain Casebooks
Realistic healthcare, finance, and enterprise cases with risk boundaries and evals.
The mental model
An agent is not magic. It is context, tools, memory, workflow, evaluation, and human approval arranged around a useful task.
Run locally
Verify Examples
Run every dependency-free example and showcase with one command.
Open scriptMini RAG
Inspect retrieval, grounded answering, no-answer behavior, and RAG evaluation.
Open exampleEvaluation Harness
Run regression checks that separate correctness, format, and safety behavior.
Open exampleBenchmark Runner
Check tool use, RAG grounding, approval gates, injection defense, and traces.
Open benchmarksGraph Approval Agent
Inspect explicit graph transitions, high-risk routing, and approval gates.
Open exampleObservable Agent
Trace decisions, tool calls, guardrails, and replayable incident evidence.
Open exampleInjection Defense
Block unsafe instructions from retrieved content before they reach the agent.
Open exampleCost-Aware Agent
Route tasks by budget, latency target, and required answer quality.
Open exampleDurable Workflow
Checkpoint work, resume safely, and preserve artifacts after interruption.
Open exampleModern MCP Gateway
Practice tools, resources, prompts, authorization, and elicitation.
Open exampleMemory Governance
Redact PII, merge repeated memories, decay confidence, and delete sensitive records.
Open exampleAgent Permissions
Assign owners, scopes, risk tiers, access reviews, and audit logs to agents.
Open exampleAdvanced Eval
Run regression, safety, adversarial, and golden trace checks before release.
Open exampleIncident Playbook
Use traces, containment, hotfix evals, and postmortems when an agent fails.
Open playbookProduct UX
Design approval, evidence, control, recovery, and trust patterns for agent products.
Open checklistOperating Model
Register, review, monitor, and retire enterprise agents with clear ownership.
Open checklistRelease Kit
Use release, v1 readiness, and deployment review checklists before publishing changes.
Open release kitGovernance Templates
Start with registry, risk assessment, and deployment review templates.
Open templatesPaper Roadmap
Study agent papers from OpenAI, Google, Meta, Anthropic, Microsoft-adjacent ecosystems, Stanford, Princeton, and Tsinghua.
Open papersPaper Notes
Read concise engineering takeaways for ReAct, Toolformer, WebGPT, RAG, Reflexion, Voyager, AgentBench, SWE-agent, and safety papers.
Open notesDeepEval And RAGAS
Compare practical frameworks for LLM app tests, RAG metrics, safety checks, regression gates, and CI evaluation.
Open guideOpen Source Map
Explore agent frameworks, MCP projects, RAG tools, eval systems, observability, and ops infrastructure.
Open projectsFramework Matrix
Choose agent frameworks by control flow, state, tools, evaluation, observability, safety, and operations.
Open matrixRepo Reading Guide
Use a 30-minute reading loop to extract architecture lessons from open-source agent repositories.
Open guidePortfolio Projects
Build GitHub-ready agent projects with evals, traces, safety boundaries, and architecture writeups.
Open projectsCapstone Starter
Start the final project from a runnable colony scaffold with regression evals.
Open starterShowcase demos
Healthcare Colony
Routes health education requests with safety boundaries and escalation rules.
Open demoFinance Research
Compares companies as research support without personalized investment advice.
Open demoEnterprise Support
Classifies support tickets, routes work, and identifies approval needs.
Open demoDomain Casebooks
Study realistic safety and evaluation cases for healthcare, finance, and enterprise agents.
Open casebooks