Build a knowledge product

This tutorial builds one knowledge product from raw material to an exported bundle. You direct a staged agent pipeline through its gates: sighting the corpus, researching sections, working each section to final in the scope workshop, and packaging a two-component product — a dense agent-facing knowledge set an AI executes from, plus a human-facing manual (PDF) a person reads.

The scenario: you have accumulated a corpus on a subject you know well — research reports, transcripts, notes — and you want to turn it into a product: a playbook a customer's AI agent can execute, paired with the manual their human reads. The harvest and analysis walks (From harvest to publish) are one good way such a corpus accumulates.

The pipeline is operator-directed by design: agents do the reading, extraction, and writing; a local engine does everything deterministic; and nothing advances because a timer fired — every stage boundary is your dispatch.

Prerequisites

• Ultra tier — Products is an Ultra surface; on Pro the view is visible with a lock (see Tiers: Pro & Ultra).
• A seed corpus: a set of raw-material files in your workspace. A handful of substantial documents is enough to walk the pipeline.
• A thesis: one or two sentences on what this product argues and for whom. Vague theses produce vague products — sharpen it before step 1.
• Your Claude subscription connected — every dispatch below is an agent run on it; the deterministic engine work between dispatches costs nothing (see BYO Claude & keys).
• Optional: a Gemini API key for the cover image and section infographics. The pipeline runs end-to-end without it.

Step 1 — Create the product

Open Products and click New in the product sidebar.

1. Name the product.
2. Write the seed thesis — what it argues and for whom.
3. Pick the seed files — the raw material the pipeline works from.

You should see: the product in the sidebar, with the Research tab now acting as its dispatch surface. (Later, an existing product can seed a new one: Use as template in the Registry pre-fills thesis, outline, and seed files.)

Step 2 — Sight the material

In the Research tab, run the dispatch cards top to bottom. First: Sight.

The sighter reads every seed file and produces a structured topic index — what material exists, where, about what.

You should see: the topic index land as the product's map. Everything downstream navigates by this map instead of re-reading the corpus — which is what keeps later dispatches focused and affordable.

Step 3 — Compose the outline

Dispatch Compose. The composition scout reads the sighted material plus your thesis and proposes a foundation outline.

You should see: the first run populate the working outline. Read it critically — this is a review point, and the outline is yours to adjust. Later runs accumulate alternatives you can adopt rather than overwriting.

Step 4 — Research dispatches

Dispatch focused research passes, one at a time. Each run targets one of three scopes:

• a single section — deepen one part of the outline,
• a general cross-section pass — material that serves the whole,
• the outstanding-research backlog — the gap ledger (below).

Each pass produces a build-ready, de-duplicated research note in the product's own voice. Researchers also maintain a shared gap ledger at the product root — what is still missing stays visible, per section, and each gap is dispatchable.

You should see: research notes accumulate per section, and the gap ledger shrink as you work it. Re-run any scope as the material grows; dispatches accumulate artifacts rather than overwriting.

Step 5 — Generate alpha (the gate into the workshop)

When the notes cover the outline, dispatch Generate alpha. The engine stages canonical per-section files mechanically, then an auditor writes a consolidated product overview — thesis, goals, cross-section ownership, duplication findings.

You should see: the overview plus the full outline — and the Scope tab open for business. Read the overview before proceeding; it is the cross-section map every later stage applies.

Step 6 — The scope workshop

The Scope tab is the human-in-the-loop heart of the pipeline. It lists every section with its current artifact state. Work the sections to final:

1. Refine a section: dispatch the refiner with the section's research note, any reference files you select, and your instructions. It transforms notes about the section into the near-final section itself — and writes a new revision each time, never editing in place, so nothing is lost to an over-eager rewrite.
2. Finalize the section: the finalizer applies the auditor's cross-section ownership map, scrubs off-voice wording, and produces the final agent-facing section.
3. Close gaps as they surface: sections flagged as thin get a targeted research pass without leaving the tab.
4. Re-architect from here if the product's shape needs to change mid-flight — the outline is not frozen.

You should see: each section's state advance as you work it. When the section set is done, Roll over — this stages both components into the draft area for packaging.

Step 7 — Package and export

By the time the product reaches the Drafts tab, the writing is done; this stage packages:

1. Review the deliverables — the agent-facing set (read-only) and the human-facing per-section documents (editable, with inline image previews).
2. Final touch-up — re-dispatch the human-facing author per section to add images or refine wording before the PDF is built.
3. Cover image — generate the product cover (this is the step that uses the optional Gemini key).
4. Package & Export — build the human-facing PDF, and optionally export the agent-facing set as MDX or PDF alongside it.

You should see: the finished bundle as ordinary files in your workspace. The export step assembles; it does not hide anything.

Where everything landed

Where to go deeper

• Products — the knowledge-product pipeline — every tab and dispatch in reference form
• Tiers: Pro & Ultra — where Products sits in the subscription
• Workspace & documents — where all of these files live
• The agent roster — the pipeline's specialist agents