Show HN: Reels-style shopping feed with a client-side recommender

Thodasa launched a short-video shopping feed that surfaces products through a client-side recommender, according to a Show HN post by author bhaumiktandan on August 22, 2026. The product is live at thodasa.com.

The core proposition is a vertically scrolling, Reels-style interface for browsing and buying products, where the recommendation logic runs in the user's browser rather than on a remote server. That architecture means user preference signals — swipes, watch time, interactions — are processed locally without being sent to a backend for personalization.

No founding team details, funding status, company size, or headquarters location were disclosed in the HN submission. The post received 1 point and no comments at the time of indexing, and no additional sources corroborate or expand on the launch details.

The short-video commerce format has become crowded, with TikTok Shop, Instagram's shopping features, and a range of live-commerce platforms competing for the same behavior. Thodasa's stated differentiator is the on-device recommender, which sidesteps both the latency of server round-trips and the data-collection surface that centralized recommendation systems typically require. Whether that trade-off — privacy and speed against the scale advantages of server-side models trained on large datasets — resonates with shoppers or merchants is an open question the sources do not address.

The HN thread contained no comments, so there is no public technical feedback on the recommender's implementation, cold-start handling, or catalog integration at the time of writing. No product roadmap, expansion plans, or merchant partnerships were mentioned in the submission.