Local Recommendation Agent: Places + Details in One Prompt
Build a local recommendation flow with an AI agent — search places by city and category, pull details and ratings, then rank by your preferences, all from one plain-English prompt.

Tool names in this guide are for reference only and may have changed or been renamed. Please use the latest tools in the tool marketplace when implementing.
Key takeaways
- Run a full search-details-rank recommendation loop from one sentence
- Use map place tools at home and a Serper places tool abroad, same prompt
- Get a ranked shortlist with reasons, not a dump of raw search hits
Tools used
Why local recommendations are a chore
"Find me a few highly rated cafes downtown in some city that are good for working" — it sounds simple, but doing it by hand is fiddly. You search the category in a maps app and get a long list of candidates, open each one to check hours, rating, price and vibe tags, then mentally re-rank them against what you actually care about (quiet? power outlets? near transit?). Change the city or the category and you repeat the whole dance.
The pain isn't missing data — it's that the data is scattered, you keep switching views, and you still have to do the final judgment and ranking yourself. An AI agent collapses the loop: you describe the city, category and preferences in one sentence, and the agent searches places, pulls details, ranks by your preferences, and hands back a shortlist with reasons.
The workflow in four steps
- Give preferences — tell the agent the city, the category, and what you actually care about (rating, price, vibe, distance).
- Search candidate places — pull a batch of candidates by city and category.
- Pull details and ratings — enrich each candidate with rating, hours and other details.
- Rank and deliver — the agent ranks against your preferences and returns a shortlist with reasons.
How to run each step
Step 1 — Describe the job
Tell the agent plainly: use ae to recommend highly rated, quiet cafes good for working in downtown example city — give me the top five with reasons. The agent interprets this and picks a place-search tool. No maps API wiring on your side.
Step 2 — Search candidate places
At home, the agent calls a place text-search tool (such as ae_amap_maps_get_v5_place_text) to pull candidate places by city and category keyword. You hand it the city plus category once instead of searching entries one by one.
Step 3 — Pull details and ratings
With candidates in hand, the agent calls a place-detail tool (such as ae_amap_maps_get_v5_place_detail) to fill in rating, opening hours, address and other details for each one. This step is what makes the recommendation grounded rather than a guess from names alone.
Step 4 — Places abroad and ranked output
If you ask about a city overseas, the agent switches to an international places tool (such as ae_serper_places) to do the same search-and-detail step. Once the data is in, it ranks by the preferences you gave up front (rating, quietness, distance) and returns a shortlist with reasons instead of raw hits.
Checking reviews by hand vs. asking an agent
| Step | Browsing maps / reviews yourself | With an AI agent |
|---|---|---|
| Find candidates | Search category, scroll a long list | One prompt pulls candidates by city + category |
| Check details | Open each one for rating/hours | Ratings and details filled in automatically |
| Home vs. abroad | Switch apps and habits | Same prompt, auto-switches place tools |
| Ranking | Weigh it in your head | Ranked by the preferences you gave |
| Output | A pile of open tabs | A shortlist with reasons |
How to get started
- Sign up — new users get 1,000 free credits.
- Connect in-app inside an agent like WorkBuddy or Trae, or paste a one-line MCP config into Claude Code, Cursor or VSCode.
- Give the instruction in natural language and let the agent search, enrich and rank.
FAQ
Does it work both at home and abroad? Yes. Domestic cities go through a map place tool, and for overseas cities the agent switches to an international places tool (such as ae_serper_places) — you just name the city in the prompt.
Can it rank by my preferences? Yes. State what you care about (rating, price, vibe, distance) in one sentence and the agent ranks accordingly and explains each pick, rather than returning a default order.
Do the recommendations include ratings? Yes. The agent uses a place-detail tool to fill in ratings, hours and other details, then ranks on that basis, so every pick is grounded.
