Procurement agents

Agents that compare and monitor vendors run on the same dated record a human reviewer would use, so the shortlist can be re-derived and defended later.

Request access

Every recommendation traces to a dated capture, not to a chat log.

The job

An agent's shortlist is only as auditable as its inputs.

Point an agent at the open web and it reads vendor marketing, then produces a recommendation shaped by whoever wrote the best landing page. Ask six months later why that vendor won and the reasoning is gone, along with the pages it was based on.

Against the record the agent works from typed facts with dates: what a category contains, what each vendor states, what was observed, and what has moved since. The output can be regenerated, and the evidence is still there when the review happens.

What that costs today

  • Shortlists whose reasoning cannot be reconstructed
  • Vendor marketing read as though it were evidence
  • Nothing watching the vendors already under contract
  • Comparisons that were current on the day they ran, only

The API

5 calls cover the job.

Current state is open. History values, claim checks, and evidence trails sit behind a free key. These are the real routes.

GETopen
/v1/categories/{slug}

The bounded candidate set, with a per-member rollup.

GETopen
/v1/companies/{company}

Current state per candidate: pricing, security, stack.

GETfree key
/v1/companies/{company}/claims

Each vendor claim against what was observed.

GETfree key
/v1/companies/{company}/trajectory

Direction over time rather than a single snapshot.

GETmasked · key unmasks
/v1/changes

What moved across the shortlist since the last run.

For agents · MCP

Or let your agent run it.

Connect the Bixel MCP server and any agent, Claude, an internal copilot, a procurement bot, works this job against live structured data, with every value citing the dated capture it came from. The recommendation becomes auditable, not a guess.

See the MCP tool list

You ask your agent

“Shortlist the vector databases on security posture and pricing, then tell me what changed since we last looked.”

The agent that shortlists needs a memory.

The category page is the candidate set an agent would start from: see the candidate set →

More in AI & Agents

The answer layer for agents that need company facts with receipts.