Training
AI Automation: Agents & Workflows
Turn repetitive knowledge work into supervised automation: LLM-powered workflows, tool-calling agents, integration with your existing systems, and the guardrails that make automation trustworthy.
Between "chatbot" and "AGI fantasy" sits the actually-valuable layer of AI: automating the repetitive 40% of knowledge work — document processing, data extraction, triage, drafting, cross-system glue — with humans supervising the judgment calls. This program is about building that layer responsibly: workflow design, LLM tool-calling against your real APIs and databases, orchestration (from n8n-style platforms to code-first frameworks), and the guardrail engineering — validation, confidence routing, human-in-the-loop checkpoints, audit logs — that separates automation you can trust from automation you'll apologize for.
Practical from hour one: participants automate a real process from their own work during the course.
Curriculum
-
1
Automation triage: which processes are AI-shaped and which aren't
-
2
LLM building blocks: structured extraction, classification, drafting
-
3
Tool calling and agents: connecting models to APIs, databases and files
-
4
Orchestration options: visual platforms vs code-first, chosen honestly
-
5
Guardrails: validation, confidence thresholds, human-in-the-loop design
-
6
Reliability: retries, idempotency, monitoring and audit trails
-
7
Capstone: one real process from your work, automated and supervised
