# AI Resolve **AI Resolve** is the operator action that fills in catalog detail from a starting point. Give it a coordinate, an address, an existing record, or a paste of identifying text, and it proposes the structured catalog data — place hierarchy, nearby landmarks, suggested attributes — and you confirm. On a typical property, the resolve gets you 80% of the way to a complete catalog entry; you spend the saved time on the actual differentiators. ## What it does You give AI Resolve something concrete: * A coordinate pin on the map. * An address or property name typed into the resolve field. * A pasted block of text describing the property (a partner's catalog entry, a website description). It returns a proposed structured record: * **Place hierarchy** — country, region, city, district, neighborhood resolved from the coordinate or address. * **Nearby landmarks** — beaches, transit hubs, points of interest within a configurable radius. * **Suggested categories** — based on the description's keywords (boutique, all-inclusive, family-friendly, etc.). * **Suggested attributes** — amenities the description implies (pool, gym, spa, restaurant). * **Multilingual place names** — the canonical labels for the resolved hierarchy in the languages your organization operates in. You review the preview, accept it as a whole, refine specific fields, or reject and try again with a different starting point. ## When you use it A few common situations: * **Onboarding a new property fast.** Drop a pin, AI Resolve fills in the hierarchy and the obvious nearby landmarks within seconds; you spend the saved time on the property's actual differentiators. * **Filling gaps in an existing property.** A property has coordinates but no neighborhood; AI Resolve proposes one. Or a property has a description but no categories; AI Resolve suggests them. * **Sanity-checking a manual entry.** You set the place hierarchy by hand; running AI Resolve against the same coordinate confirms it (or flags a disagreement worth checking). * **Bulk-onboarding from a content feed.** The feed gave you names and addresses for 60 new properties; AI Resolve fills in the hierarchy, landmarks, and category suggestions for the whole batch in one pass. ## What it does not do AI Resolve is a **proposal**, not an oracle: * It does not commit anything until you confirm. * It does not override existing values you have set unless you explicitly accept that. * It flags uncertainty — a coordinate in a sparse rural area, an ambiguous neighborhood boundary, or a description in a language with weaker grounding. * It does not replace local knowledge. A new property in a niche neighborhood might have specifics AI Resolve cannot know; the resolve gets you 80% of the way, you do the last 20%. ## How it appears in the property edit flow When you are creating or editing a property, AI Resolve shows up in two places: * **On the location step** — the *Resolve* button next to the coordinate field. Click it; the place hierarchy and nearby landmarks panel fills in with proposed values; you confirm or adjust. * **On the content step** — the *Suggest from description* helper. Paste a block of text, AI Resolve proposes categories and attributes; you select which ones to apply. The results are previewed before commit, the same way every Adragent operation is. Cell history records the resolve as the source of any field it filled in. ## How it relates to Adragent Adragent uses the same machinery when you ask: * *"Resolve the location for the new Hotel Sunrise in Antalya at 36.85, 30.78."* * *"Find every property without a neighborhood and resolve it."* * *"Suggest categories for the Marina property based on its current description."* The AI Resolve action is the operator-facing surface; Adragent is the conversational surface; both reach the same proposal engine. ## Provenance Anything AI Resolve fills in carries provenance — the field's origin is recorded as *AI-resolved, accepted by Operator A on date D*. Two consequences: * You can later trace exactly which fields you accepted from a resolve and which you set yourself. * If you decide AI Resolve made a wrong call on a class of properties, you can find them, override, and remove the resolved-source flag. This is the same source model used elsewhere in the catalog (see [Canonical mapping](/console/mapping-and-curation/canonical-mapping.md)) — AI Resolve is one more source contributing suggested values. ## Where to next * **The location data AI Resolve fills in** → [Location & geography](/console/catalog/location-geography.md) * **The categories it suggests** → [Categories](/console/catalog/categories.md) * **The attributes it proposes** → [Rooms & attributes](/console/catalog/rooms-and-attributes.md) * **The Adragent surface that drives the same machinery** → [Adragent overview](/console/adragent/overview.md) --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://adrasis.gitbook.io/console/catalog/ai-resolve.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.