Agentic applications
Un sistema que puede seguir en marcha. Memoria, herramientas, un aislamiento, un horario, evaluación y un límite claro de lo que puede hacer.
19 partes · ~102 h
EmpezarLecciones
- 00Course 2: agentic applications on local models — index
- 01Part 1 · Agent architectures — a map of the terrain
- 02Part 2 · The skeleton — process, configuration, sessions
- 03Part 3 · Harness — a loop that survives long tasks
- 04Part 4 · Context engineering — the attention budget
- 05Part 5 · Long-term memory — three schools
- 06Part 6 · Identity — working files as the program
- 07Part 7 · Tools — design and ergonomics
- 08Part 8 · Code-mode — the agent writes a program instead of calling tools
- 09Part 9 · Protocols and integrations — MCP, CLI, A2A
- 10Part 10 · The production sandbox
- 11Part 11 · Scheduling — an agent that wakes itself up
- 12Part 12 · Channels — how you talk to the agent
- 13Part 13 · The fleet — many agents, one system
- 14Part 14 · Evaluation — how you know this works
- 15Part 15 · Observability — seeing what it does
- 16Part 16 · Agentic security — OWASP ASI01–ASI10
- 17Part 17 · Hybrid and router — local by default, cloud by conscious choice
- 18Part 18 · Release and maintenance
- 19Part 19 · Capstone — a general-purpose personal assistant
- —Sources — what this course stands on