AI Image / Visual Search

Because this is build-to-order, ECOSIRE builds the module against your exact Magento version, theme, and catalog before installing. Typical delivery is roughly 2 to 4 weeks depending on catalog size, embedding model choice, and integration depth (standard Luma/theme vs. headless PWA via GraphQL). We confirm a firm timeline after a short scoping call, install on staging first, then promote to production.

Every build includes a post-launch support window for bug fixes and compatibility with Magento minor releases. Because the module is isolated under app/code/Ecosire/VisualSearch with no core edits, Magento patches and most upgrades apply cleanly. ECOSIRE offers continued support and feature work after the initial window; we keep the service contract and admin config stable so updates don't break your storefront.

Yes. The module is built on standard Magento 2 architecture (service contracts, di.xml, plugins, observers, ACL) so it runs on both Magento Open Source and Adobe Commerce. On Adobe Commerce we can align the visual results with Live Search/Catalog Service where it adds value, but the core visual matching does not depend on Commerce-only features.

We scope this with you. The embedding step can call an external AI provider (vectors only, not raw customer data) or run against a self-hosted model, depending on your data-residency and cost requirements. Product images are embedded during a cron-driven background job; queries embed the uploaded image and match it against your local vector index. We document exactly what is sent where before build.

No. Image embedding runs in a cron-driven background indexing job, not inline with catalog saves. An observer on catalog_product_save_after only flags changed products for re-embedding. At query time the storefront calls the service contract, which performs a fast nearest-neighbor lookup against the prebuilt index, so visual search adds minimal latency to the page.