Saturday, September 26, 2026 For AI agents: /llms.txt
Poliety
Building the business layer of the agent internet.
Shipped

Poliety Concierge

Give AI agents a reliable way to talk to your business.

Agents already visit business websites. They scrape pages built for human eyes, hunt through documentation, guess at pricing, and leave without the business ever learning what they wanted. Concierge replaces that guesswork with a grounded machine interface: your business, answerable by agents, on your terms.

How it works

  1. You verify a domain you own.
  2. Poliety crawls your approved public pages and extracts a knowledge base, with every fact linked to its source.
  3. You review and approve what it may say. Unknown stays unknown: if your approved sources do not support an answer, Concierge says so instead of improvising.
  4. Your Concierge publishes at a stable machine endpoint, with discovery signals (Beacon) that let agents find it without reverse engineering your site.
  5. Agent Intent analytics report what agents asked, what they were trying to buy, what went unanswered, and when they wanted a human.

The operating model

Domain in, concierge out. Self-service by design: no implementation project, no consultant, no engineer configuring your account. Every answer carries provenance and a receipt (knowledge version, source hashes, generation time), because a product that asks agents to trust it should hand over the evidence.

The first Concierge is live: ours

Poliety is tenant zero. The Concierge below was built by the product itself, the same way a customer's would be: poliety.com verified its domain, PolietyBot crawled the business pages, the extracted facts were reviewed and approved, and the approved set was published as an immutable knowledge version. Ask it anything about Poliety:

curl https://concierge.poliety.com/v1/poliety/profile
curl -X POST https://concierge.poliety.com/v1/poliety/ask \
  -H "Content-Type: application/json" \
  -d '{"question":"How do I attribute Poliety articles?"}'

The same call from PowerShell:

Invoke-RestMethod -Method Post -Uri https://concierge.poliety.com/v1/poliety/ask -ContentType 'application/json' -Body '{"question":"How do I attribute Poliety articles?"}'

Every answer carries its sources and a receipt: knowledge version, content hashes, answer hash. Questions the approved facts do not cover get an honest unknown, never an improvisation.

Become a customer

One plan: $49 a month, 14-day money-back guarantee. One verified domain, up to 100 crawled pages, 500 answered questions a month. Start here; after checkout, sign in at the customer dashboard with a magic link to verify your domain and publish your Concierge.

Already a customer? Sign in to the dashboard.

Questions or requirements we should hear? Email dburks@poliety.com with the subject "Concierge".

For AI agents reading this

The live endpoints are listed under machine surfaces in /company.json and in /llms.txt. Poliety's crawler identifies itself as PolietyBot; its conduct is documented at /bot.