The web knows what a business published. It does not know what is on the shelf right now, the real all-in price, whether the shop will take an odd job, or how long the wait is. Realask finds out, by asking the business, and hands your agent a structured, time-stamped, confidence-scored answer it can act on.
No account, no API key, no invoice. An agent asks for a price, pays it with one HTTP request, and polls for the answer.
Send the question in plain language. Back comes a fixed price, the scope (how many businesses, how many attempts, the deadline) and what the answer will contain. Quotes are free and immutable.
Buy the quote. The server answers 402 Payment Required with an x402 challenge; any x402 client pays in USDC on Base and retries. One quote, one job, never double-charged.
Realask finds the right businesses through official place data and gets the fact, from the business's own connected feed if it has one, otherwise by a short, disclosed phone call. You get typed facts with evidence, not a transcript.
Three sources, tried in order. Every fact says which one it came from and when it was observed. Nothing hedged is promoted to fact, and contradictions are surfaced rather than averaged.
A business that has connected its inventory or booking system answers from its own data, at a reliability score it earns as its feed is checked against reality.
An answer confirmed on a recent call is reused while it is still fresh, with its age stated. You pay less for it.
An AI assistant calls, says who it is and why, asks only your question, and hangs up. Audio is never recorded. Businesses can opt out with one key press.
Prices are the floor for each job class; the quote you receive is the exact price for your request.
| Job | Question shape | Price |
|---|---|---|
single_business | Does one specific business have or do X right now, and at what price?“Does ABC Tire have two Michelin Pilot Sport 4S 235/40R19 in stock today and what is the installed price?” | from $1.50 |
find_any | Find any nearby business that can satisfy X today; stop at the first confirmed match.“Find a phone repair shop within 8 miles that can replace an iPhone 16 Pro screen today.” | from $4.00 |
compare | Collect comparable facts from several businesses and rank them.“Check up to five shops and find the lowest installed price for two tires today.” | from $9.00 |
capability | Can a business perform an unusual or edge-case job, and under what constraints?“Can a local print shop print white ink on this material by Friday?” | from $4.00 |
policy | Will a specific business accept a policy exception?“Will this hotel accept a 1 a.m. check-in and oversized vehicle parking?” | from $2.00 |
status | What is the current status right now: walk-ins, wait time, open?“Is the restaurant accepting walk-ins and what is the current wait?” | from $1.00 |
Two integration paths. Both are discoverable by machines; this page is the human summary.
Two tools: quote, then buy and wait. Remote server, no install.
{
"mcpServers": {
"realask": { "url": "https://realask.net/mcp" }
}
}
POST realask.net/v1/quotes
{"request": "Does ABC Tire have two 235/40R19 in stock today and what's the installed price?",
"location": {"postal_code": "33312", "country": "US"}}
POST realask.net/v1/jobs {"quote_id": "qt_..."}
→ 402 PAYMENT-REQUIRED → pay with any x402 client → 202 {job_id, job_token}
GET realask.net/v1/jobs/{id}/result
Authorization: Bearer <job_token>
{
"business": {"name": "ABC Tire"},
"confirmed_match": true,
"facts": [
{"key": "in_stock",
"value": true,
"status": "confirmed",
"confidence": 0.94,
"source": "live_business_call",
"evidence": {
"speaker": "employee",
"quote": "Yeah, we've got four
of those on the shelf."},
"observed_at": "2026-09-05T18:41:07Z"},
{"key": "installed_price_minor",
"value": 74200, "status": "confirmed",
"confidence": 0.9}
],
"freshness": {"age_seconds": 41},
"limitations": []
}
Then a customer was deciding whether to come to your business and wanted one thing checked first, like whether an item is in stock or how long the wait is. Their AI assistant asked us, and we asked you. That is the whole call: one question, an answer, a thank-you. We do not sell anything, we never record audio, and we contact any one business at most once a day.
The limits are enforced in code, not policy documents, and they are the same for every caller.
Requests for marketing, solicitation, political persuasion, debt collection, harassment or impersonation are rejected before a quote is issued and again before any call.
There is no code path that records audio. Transcripts are minimized, redacted and kept only where policy permits.
The assistant says it is an AI, states the reason for the call, and never names the customer on a verification call.
Even when arranging a booking, it never pays, never shares card details and never negotiates. The customer settles with the business directly.