LESSON 06 / 16·2 min readIntermediate

Workflow State and the Ledger

Must knowConcept lesson

A multi-turn workflow needs a reliable memory of what has happened. That is different from merely sending the whole chat history back to an LLM.

A workflow ledger can store structured, validated facts such as:

{
  "customer_verified": true,
  "transaction_id": "txn_example_123",
  "dispute_reason": "unrecognized_transaction",
  "replacement_card_requested": null,
  "action_confirmed": false
}

The ledger should distinguish between:

Unknown: No usable answer has been collected yet.
Known: A value has been collected and validated.
Pending: A required question or confirmation is still open.
Completed: A permitted action or workflow step has finished.

This prevents the system from treating conversational wording as authoritative state.

PAUSE & RECALL

Can you explain it without the notes?

Workflow State and the Ledger

Use your own words. Explain an example and one limitation before checking the explanation.

Check your explanation

A multi-turn workflow needs a reliable memory of what has happened. That is different from merely sending the whole chat history back to an LLM.

CONNECT IT TO YOUR INTERVIEW

Turn this concept into a clear answer.

Workflow State and the Ledger

Deterministic AI Workflows·State, rules & graph execution

A multi-turn workflow needs a reliable memory of what has happened. That is different from merely sending the whole chat history back to an LLM.

Intermediate2 min read

Practice this concept

YOUR NEXT STEP

Keep connecting the ideas.

Next lesson: Graphs, Nodes, and Transitions

Adapted from 19_deterministic_workflow_engine_learnings.md · “5. Workflow State and the Ledger”. Examples are study fragments; framework behavior can depend on your version.