The Bear House says Glance AI lifted conversion 5X and AOV 75%
The apparel retailer reported more than 3X higher site time and 20% lower returns after deploying Glance’s agentic-commerce tools. Separately, Manyavar is using two decades of data to guide assortment and store-stock decisions, underscoring retail’s shift toward vertical AI models.
What happened
Indian firms including Razorpay, Fractal and BharatGen are developing domain-specific AI models. Retail-relevant examples show The Bear House using agentic
Key facts
- Razorpay transaction decisions can involve tens of thousands of structured signals
- The Bear House reported a 5X conversion increase, over 3X more site time and 20% lower returns using Glance agentic commerce
- The Bear House average order value rose from ₹2,000 to ₹3,500
- Manyavar uses nearly 100 Cr data points from two decades of operations
- Adapting a 7-13B model is estimated at ₹40 Lakh-₹3 Cr over 3-6 months
- Sector-specific model training is estimated at ₹8-15 Cr; building a large model from scratch at ₹80-150 Cr
- Annual model updates, testing and compliance may cost ₹5-8 Cr
- IndiaAI subsidies and lower engineering costs could make builds 30-50% cheaper in India than the US
Why this matters
Glance’s traction and Manyavar’s data-led assortment strategy highlight partnership and acquisition opportunities in vertical AI platforms that connect customer discovery, merchandising and inventory decisions.
What to watch
- Independent or cohort-level validation of The Bear House's 5X conversion, 75% AOV and 20% returns-reduction claims.
- Whether gains persist after novelty fades and expand to repeat customers, mobile traffic and lower-intent acquisition channels.
- Changes in gross margin, markdown rate and return-processing costs, not just revenue per visit.
- Manyavar or peers disclosing inventory-turn, stockout or sell-through improvements from vertical assortment models.
- Major commerce platforms adding native AI styling, fit and merchandising agents that commoditize point solutions.
- Consumer feedback or regulatory scrutiny around AI-generated recommendations, profiling and data consent.
- Deploy AI agents first on high-intent apparel journeys: size/fit guidance, outfit bundling, product discovery and cart recovery.
- Measure incremental gross margin per session, return-adjusted AOV, conversion by traffic source, and customer-service deflection rather than headline conversion alone.
- Connect conversational and visual-shopping data to merchandising systems so demand signals inform buys, replenishment and store allocation.
- Create governance for recommendation accuracy, discount leakage, customer-data use and hallucinated product or policy claims.
- Expect apparel platforms and marketplaces to package agentic-shopping capabilities, raising the cost of remaining on static catalog and search experiences.
Also reported by
- Inc42 — Same time