Deep Research API

Research, with Verified Addresses

One endpoint. You ask a question about a place; we return a deep research report where every recommended venue is cross-checked against verified location data for real address and walking distance from your anchor.

Read API Docs
What's Different

Web Research Hallucinates Addresses. We Don't.

Web-only research

Returns a polished markdown report. Cites web sources. But the addresses and operating hours come from the same web pages that drift over time — closed venues, stale phone numbers, chain locations in the wrong city.

You don't know which claims are real.

Camino /report

Same web-research-quality prose, plus every cited venue is cross-checked against verified location data. Wrong-city collisions drop into drift_warnings. Confirmed places get the real address and walking distance from your anchor.

Every recommendation is ground-truthed.

Real Examples

Three Reports, Verified Out Loud

Below are real outputs from /report. Every address shown was ground-truth verified and is annotated with the walking distance from the user's anchor coordinate.

Best craft breweries in NoDa Charlotte
35.247, -80.811 · 2km radius
Verified Places
  • NoDa Brewing Company brewery 2921 N Tryon St — 508m from anchor
  • Heist Brewery brewery 2909 N Davidson St — 232m from anchor
  • The Chamber by Wooden Robot tap room 416 E 36th St — 517m from anchor
What the synthesizer wrote

Three real NoDa taprooms picked from 11 web sources. Wooden Robot has two Charlotte locations; the pipeline correctly disambiguated to The Chamber (the NoDa taproom on 36th St) rather than the South End original 5km away.

No drift warnings — the research stayed within the requested radius.

Best brunch and specialty coffee in Amsterdam Jordaan
52.376, 4.883 · 1.2km radius
Verified Places
  • Kafenion coffee shop Bloemgracht — 183m from anchor
  • Box Sociaal brunch Bloemgracht 47 — 178m from anchor
  • Winkel 43 cafe Noordermarkt 43 — 407m from anchor
  • Greenwoods Singel brunch Singel 103-H — 560m from anchor
What got caught and dropped
drift_warnings (3)
  • Moods Coffee & Brunch — 6358km from anchor (same-name venue in another country)
  • Bruno's — 213km from anchor (likely different place)
  • Amsterdam Central Station — 1.2km, outside requested radius

Web search returned a global same-name match for "Moods" and "Bruno's"; the pipeline dropped them rather than render their addresses as "verified." The synthesized report describes only the four actually-in-Jordaan venues.

Best coworking in Lisbon Chiado for US-timezone remote workers
38.71, -9.143 · 1.5km radius
Verified Places
  • Heden Chiado coworking Largo Rafael Bordalo Pinheiro 29 — 227m
  • Impact Hub Baixa Chiado coworking Tv. das Pedras Negras 1 — 609m
Synthesis weaves addresses inline

Generated markdown reads:

Heden Chiado — Largo Rafael Bordalo Pinheiro 29, 2nd floor, 1200‑369 Lisboa (227m from your anchor) — 24h member access, 200 Mbps fiber, dedicated phone booths. Hot-desk €249/month [1].

The address and distance come from our verified location data, not the web sources. Citations like [1] still link to the original web pages for verification.

Beyond Consumer Queries

Built for B2B research too

The same verify-then-synthesize pipeline that powers a brunch report handles site selection, regulatory filings, and per-account sales prep. Same API; far better unit economics at scale.

Real estate & site selection

"Coworking density and amenity profile within 500m of {candidate location} — verified address, walking distance, category breakdown."

Field sales preparation

"Best meeting venues near {prospect HQ} matched to their brand vibe" — generated per-account, no human research time.

Logistics & fleet ops

"Driver-friendly stops within 3km of {warehouse}" — restrooms, parking, food open at the right hours, addresses confirmed.

Insurance underwriting

"Commercial property due diligence on {address}" — what's actually around, with verified-not-hallucinated addresses for filings.

Corporate travel

"City briefing for {employee traveling to Tokyo}" — neighborhoods, late-hours options, walkable food — generated per-traveler.

Healthcare network adequacy

"Specialist coverage in {zip}" — verified addresses suitable for CMS / state regulatory filings.

Competitive intelligence

"All matching businesses within 1mi of {target location}" — chain density studies with confirmed locations and category data.

M&A diligence

"Neighborhood context for {portfolio of acquired retail locations}" — batch verification with drift detection on stale data.

Talk to us about enterprise use
How It Works

Four passes, one report

1. Web Research

A research pass gathers and synthesizes sources from the web with full citations.

2. Extract Venues

An AI extractor reads the prose and lists the distinct venues being recommended (skipping orientation landmarks).

3. Verify

Each venue is cross-checked against verified location data. Out-of-radius or same-name-different-city hits land in drift_warnings.

4. Synthesize

A final pass rewrites the prose to weave the verified address and walking distance for each recommendation inline.

Async by Design

Submit, Then Poll

Reports take 60–90 seconds. POST returns a job_id immediately; poll the GET endpoint until status is complete. 

# Submit
curl -X POST "https://api.getcamino.ai/report" \
  -H "X-API-Key: YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"topic":"best ramen in Shibuya","location":{"center":{"lat":35.66,"lon":139.7},"radius_m":1500}}'

# → {"job_id":"rpt_abc...", "status":"pending", "poll_url":"..."}

# Poll
curl "https://api.getcamino.ai/report/rpt_abc..."

# Response on completion includes:
# - summary (markdown prose with inline citations)
# - verified_places (name, address, lat/lon, distance_m, category)
# - drift_warnings (places dropped — usually wrong city collisions)
# - citations (web sources from the research pass)

Ready to Build?

100 free calls per month. No credit card required.

Read API Docs