A parts search reads “front brake pads for a 2021 sedan, 2.5 engine, in Jeddah today”, checks fitment and stock across nine branches, and returns two fitting pad sets with SAR prices and branch stock, plus one set excluded by a fitment rule.
Reference parts search on a fictional catalogue (illustrative example). Fitment, price and stock come from approved tools, never from the model.

Agentic Search for Any Store

The wrong part never reaches the customer, and the nearest branch with stock is named. Agentic search lets staff and customers ask for a part the way they would ask an experienced parts advisor: front brake pads for a 2021 sedan, 2.5 engine, in Jeddah today. It reads the request in Arabic or English, asks for a missing detail when it needs one, and calls approved catalogue, fitment and stock tools. Price, compatibility and branch rules run in code, never in the model, so the answer is a short list of parts that fit, with stock per branch and a clear next step. The same pattern works for any structured catalogue. Ting designs it around your data, branches and systems. For the contracts, code path, test set and production checklist, download the full Build Guide.

Agentic search · reference flow

How it runs

  1. RequestArabic or English, in plain words
  2. Interpret & clarifyVehicle · part · branch
  3. Approved toolsCatalogue · fitment · stock
  4. Rules & rankingDeterministic code
  5. Grounded resultsParts · branch stock · next step
Price, stock, fitment and branch policy stay outside the model.