03 · Agentic AI

"The Worker." LLM + loop + tools.

An agent observes its environment, decides what to do next, calls a tool, sees the result, and repeats until the goal is met. Powerful, early-stage, governance-critical.

Observe. Think. Act. Repeat.

The ReAct loop: observe state, reason about it, take an action via a tool, read the result, decide what is next. Each loop spends tokens and increases the chance of cascading errors. Cap the loop count. Audit every tool call.

What it can reach

Tools, memory, places to earn its keep.

Tools the agent can use

  • Web search and browsing
  • Code interpreter
  • File read and write
  • API calls (REST, GraphQL)
  • Database queries
  • Calendar and email

Memory types

  • Working: context window
  • Episodic: prior conversations
  • Semantic: vector store
  • Procedural: tool schemas

Where it earns its keep

  • Multi-step research (legal, market)
  • Service desk triage
  • Document-to-action workflows
  • Coding assistants that test and run
Non-negotiable

No autonomous reach into high-stakes domains.

No autonomous agent actions in finance, HR, legal, security, or customer commitments. Human-in-the-loop on any tool call with real-world side effects. This is the BIITS rule. Write it into the policy before you write the code.

The question to ask

Which actions can it take without a human?

Anything with a financial, regulatory, or customer-visible consequence: the answer must be "none." Read-only and draft-only actions are where agentic AI earns its keep right now.

Want the boardroom version of this?

Back to AI Advisory Talk to Jo