Shopsense AI unveils commerce-focused AI model for shopping recommendations

By AI, Created 12:26 PM UTC, June 02, 2026, /AGP/ – Shopsense AI launched its Shoppable Intelligence Model on June 2, saying the proprietary multimodal system beats OpenAI’s CLIP and Google’s SigLIP2 on shopping retrieval benchmarks. The company says the model can improve product recommendations, engagement and commerce revenue for publishers without changing existing integrations.

Why it matters: - Shopsense AI is pitching SIM as a commerce-first AI model that is built to improve product discovery, not just general content recognition. - Better retrieval accuracy can translate into more relevant recommendations, higher click-through rates and stronger revenue for publishers, retailers and advertisers. - The company says the gains flow through existing storefront and in-content commerce units, so customers do not need extra integration work.

What happened: - Shopsense AI announced the Shoppable Intelligence Model, or SIM, on June 2 in San Francisco. - SIM is the company’s next-generation proprietary multimodal AI for real-time contextual commerce. - Shopsense says SIM outperforms OpenAI’s CLIP and Google’s SigLIP2 on benchmarks tied to product discovery. - The announcement was first reported by MediaPost, with a link to the original article.

The details: - Shopsense says SIM delivers 25% to 50% higher retrieval accuracy across the benchmarks it considers most important for shopping. - On Fashion200K and FashionGen, SIM outperforms open-source baselines across image-to-image, image-to-text, text-to-image and text-to-text retrieval. - Against OpenAI’s CLIP, SIM reaches up to 77% higher image-to-image retrieval accuracy and up to 74% higher text-to-text accuracy on fashion datasets. - Against Google’s SigLIP2, SIM leads by 34% to 60% across retrieval tasks. - People Inc. Chief eCommerce Officer Tory Brangham said the real benchmark for AI is whether it can monetize content, and said SIM ties intelligence directly to intent and conversion. - Shopsense says a 10 percentage-point retrieval improvement can translate into better recommendation quality. - The company says a 10% improvement in model precision produces a 24.5% improvement in shopper click-through rate for native retailer media activations. - SIM is optimized for fashion and apparel, accessories, furniture and core commerce use cases. - The model is designed for wardrobe identification, style matching, outfit-level curation, bag and jewelry identification, eyewear and footwear recognition, room-level product identification and cross-category matching.

Between the lines: - The announcement positions commerce AI as a specialized category where benchmark gains are tied to measurable revenue outcomes. - Shopsense is drawing a sharp line between general-purpose multimodal models and systems trained specifically on shopping behavior and purchase intent. - The emphasis on reproducible public benchmarks suggests the company wants buyers to evaluate SIM as a performance product, not a marketing claim. - Bryan Quinn, Shopsense AI president and co-founder, said the difference between general-purpose AI and commerce-trained AI is the difference between surfacing something a viewer glances at and something that drives action.

What’s next: - Shopsense says SIM will automatically improve performance across storefront and in-content commerce units for customers using the model. - The company says its broader commerce platform is already live with Bell Media and CTV and across web publisher and creator networks. - Shopsense says its system connects 100 million daily products across 500,000 brands and 1,000-plus retailers. - The company also points to patent-pending systems for Feed Enrichment System (#63/564,250) and AI Recommendation System for Shoppable Experiences (#63/653,081).

The bottom line: - Shopsense AI is betting that the next competitive edge in commerce AI will come from models trained to convert attention into purchases, not just identify what is in an image or video.