Contents
Agentic AI is changing how technical information gets created, checked, and released—but not every workflow should be handed to a free‑roaming “AI agent.” In this talk, I’ll show a practical decision model for documentation teams: when to use deterministic orchestration (explicit, auditable workflows) and when to embrace agentic loops (adaptive problem‑solving). I’ll compare the two styles on predictability, governance, review effort, and measurable throughput, and I’ll illustrate both with the same end‑to‑end scenario: 1) draft, 2) validate, 3) publish. You’ll see UX patterns for human‑in‑the‑loop control, guardrails that keep outputs safe, and a set of metrics used in both types of flows. You’ll leave with inspiration and a checklist you can implement next week at your own work. All this to help you plan and execute on how to move from point tools to orchestrated, AI‑assisted authoring at scale.
Takeaways
After this session, attendees can: know when to use a deterministic vs. agentic workflow, explain the trade‑offs to stakeholders, sketch a successfully orchestrated workflow including clear success metrics.