Procurement Workspace
Video transcript
00:00 — The Procurement Workspace
The Procurement Workspace. Search, curate, compare, commit.
00:08 — Welcome to the Procurement Workspace
This is where a design turns into purchasable parts. Search catalogs and the web for candidates, capture supplier quotes, let the AI compare them Pugh-style, and commit a sourcing decision that rolls into your BOM. One workspace is one purchase decision — a BOM line, or a cluster of interchangeable parts.
00:34 — Link to a hierarchy node first
Attach the workspace to a node in your System Hierarchy so the AI compare can ground its scoring in that node's cost target, requirements, and region preference. A standalone workspace still works, but it's ungrounded — Seed from node and requirement-aware spec-fit need the link. You can link now or reassign later from the Past Workspaces drawer (hamburger, top-left).
01:03 — Find candidates
Type a part number to search the electronics catalog, or a plain-English description with AI web lookup ticked to pull web leads. Turn the electronics catalog off for mechanical parts, so you don't spend catalog quota on something it cannot match. Every result lands as a candidate to investigate — a lead, not yet an offer.
01:30 — Add a candidate by hand
Found a part offline — a local stockist, a distributor quote, a marketplace listing? Add it here with its price, currency, MOQ, lead time, price basis (ex-works / landed / GST-inclusive) and supplier location. Human-entered numbers are trusted (verified), and price basis lets the AI compare normalize quotes that mix GST-inclusive and ex-GST.
02:05 — Curate the shortlist
Move each candidate through the funnel with its status dropdown — reject aggregator/junk pages (with a reason), shortlist the real options. Expand a card to record an investigated quote, run the Supplier Conversation (email + WhatsApp, supplier-visible) or loop in a teammate on the Internal thread (the supplier never sees it).
02:31 — Compare, then commit
Switch to Compare and run the AI scorer — a Pugh-style matrix over your non-rejected candidates, grounded in the node's constraints, with price/lead/MOQ scored deterministically from the real numbers. Pick the winner (or override it with a rationale), draft an RFQ, and commit — the decision produces a push-to-BOM payload.
02:59 — docs.creopus.ai
Full documentation, including every tool, is at docs.creopus.ai
Search → curate → managed RFQ → compare → push to BOM.

What it's for
A working procurement notebook — one workspace per purchase decision, holding a list of candidate parts. A three-tier search finds candidates (electronics catalog via Nexar, manual entry, or opt-in AI web lookup, each marked verified or unverified), you send managed RFQs, compare quotes in an AI matrix, and push the winner to the BOM.
When to use it
When you’re actually sourcing — after a BOM exists and needs real quotes.
How it works
- Create a workspace (“What are you buying?”), then Search or Seed from node. → A three-tier async search returns candidates (Nexar catalog / manual / opt-in web-assist), each tagged verified or “unverified lead”; a budget meter shows calls used.
- Curate candidates and set each status. → Status runs exploring → shortlisted → contacted → quoted → rejected → selected → purchased; edit details to add supplier email/phone and price/MOQ/lead.
- Draft and send a managed RFQ. → The Supplier lane drafts an RFQ email (SES) or WhatsApp message; copy it, open in your own client, or “Send now (managed)”, then record the reply. An Internal lane keeps team chat separate from suppliers.
- Switch to Compare and Score with AI. → A Pugh-style matrix scores candidates against weighted criteria with a ★ winner and “Grounded in” badges over a sortable facts table.
- Select & commit, then Push to BOM. → Commit captures a rationale and writes the verified values onto the linked BOM line (“✓ Rolled into BOM”), or gives you a copyable payload.
Mechanics
- 3-tier search (catalog Nexar / manual / web-assist) as an async job
- Verified-vs-unverified provenance on every candidate
- Managed RFQ over email (SES) + WhatsApp · two-lane Supplier/Internal messaging
- AI Compare matrix (Pugh-style) · evidence photo annotator
- Push-to-BOM cost write-back