Probe
Fixed geo-local prompts across major assistants — same wording, logged models, scheduled runs.
Local AI discovery ops
Accord
Many assistants. One accord for every location.
Measure how ChatGPT, Claude, and Perplexity recommend you — fix what they get wrong — and prove dispersion drops week over week.
Live probe
Best matcha · West Village, NYC
Synced 2m ago
Venue
GPT
PPX
Claude
Gemini
Kōhī House
1
2
1
2
Matcha & Co.
2
1
3
1
Your location
—
4
—
3
Green Ritual
3
3
2
4
Dispersion
0.74
Mention rate
38%
Fact errors
2
Probed enginesChatGPTClaudePerplexityGemini
The gap
National GEO tools miss the door.
A matcha bar in the West Village loses foot traffic when Claude omits you, Perplexity ranks you fourth, and ChatGPT lists the wrong hours. Operators need per-place visibility — not one logo on a national chart.
New York City · prompt pack
“Best matcha in the West Village”
Workflow
Measure dispersion. Fix what models read. Prove lift.
Parse & score
Ranked lists, citations, and factual claims → mention rate, dispersion, competitor share.
Fix at source
Tasks tied to directories, reviews, and listicles models actually cite — not generic SEO.
Verify
Weekly re-probes and location reports operators can act on — and investors can trust.
Outcomes
Metrics that map to foot traffic.
↓ 40%
Dispersion index
Cross-engine disagreement on rank and presence.
38%
Mention rate
Prompts where your location appears at all.
0
Fact errors
Wrong hours, address, or category — tracked to zero.
3.2
Mean rank
Average position when you land in the recommendation set.
Technology
Multi-engine measurement. Claude-powered analysis.
We probe several assistants to capture platform drift. The Anthropic Claude API structures extraction, scoring, recommendations, and weekly location reports — with evals tuned for model updates.
We're not selling another mention chart. We bring every engine into accord — lower dispersion, correct facts, for every pin on the map.
Pilot program
30 days. Measured beta.
New openings and multi-location F&B in New York City — baseline probe, fix queue, weekly re-probes, and a final dispersion report for operators and investors.
- 0140+ geo-local prompts per location
- 023–5 competitor benchmark set
- 03Source-level fix priorities
- 04Claude-generated weekly location report
Early access
Request pilot access
Share your brand and location count — we'll reply with cohort availability and onboarding steps.
Prefer email? hello@getaccord.com