Retail site selection briefs from new Airtable candidates

Build me an agent workflow that produces a full location-intelligence site evaluation brief in Google Docs every time a new candidate retail site shows up in our Airtable intake base.

Trigger: poll Airtable with the new_record event type on a 'Sites Under Review' style table. Up front, ask me for the Airtable base id, the table id or name to watch, and the Google Drive folder id where finished briefs should be saved. Assume each Airtable row carries at minimum: the property address, the target category (for example 'quick-service restaurant'), and the chain we are scouting for at this site. Default the analysis window to the trailing 12 months unless I override it.

For each new Airtable row, the agent should do the following:

1) Call Placer.ai 'Search for Placer Entities' to resolve the address into the closest matching Placer venue or complex. Prefer a venue match in the target category when one is available; fall back to the nearest complex. Capture the resolved entity id, entity type, name, and category, and note in the brief which entity was matched so the broker can sanity-check it.

2) For that resolved entity, run these Placer.ai reports against the trailing 12 month window: 'Visit Metrics Summary Report' for total visits and visitors, 'Trade Area Demographics Profile Report' for the true trade area population profile, 'Ranking Metrics for a Single Property' to get the entity's rank within its category, and 'Visitor Journey - Favorite Chains' to surface which chains the location's visitors also frequent. Remember Placer's report endpoints are async: a 202 means retry the same request body after a short delay until you get a 200.

3) Use the target chain on the Airtable row to highlight competitor co-visitation: look up where the target chain ranks in the 'Favorite Chains' results, and call this out explicitly in the brief (high overlap = strong shopper fit; low overlap = audience mismatch to flag).

4) Call Google Docs 'Create Document' to create a new doc titled something like '[Site Brief] {address} - {target chain}', then call 'Batch Update Document' to populate it with these sections: header with address, target category, target chain, resolved Placer entity, and analysis window; Visit volume vs category benchmarks (numbers from Visit Metrics Summary, framed against category context from the ranking call); Trade area demographic fit (key cuts from the Demographics Profile, called out as fit/non-fit against the target chain's typical shopper); Rank within category (position from Ranking Metrics, with a one-line interpretation); Competitor and co-visitation (top favorite chains, with the target chain's position called out); and a short 'Recommendation' summary the broker can read in 30 seconds. Move the new doc into the Drive folder I configured at setup.

5) Call Airtable 'Update Record' on the originating row to write the new doc URL back into a 'Site Brief' link field (create or reuse this field) so the broker sees the brief inline in the base. Also stamp a 'Brief generated at' timestamp field on the same row.

If Placer cannot resolve the address to any entity, or a report returns 204 (no data), write the brief anyway with the missing sections labeled clearly and still update the Airtable row, so the broker knows the agent ran but flagged gaps. Surface errors back into the Airtable row in a 'Status' field rather than silently failing.