ThredUp Deploys Agentic AI to Personalize Shopping

PYMNTS reports that ThredUp has added an "agentic" AI shopping experience to its resale marketplace, and the feature is live for a subset of customers, ThredUp CEO and Co-Founder James Reinhart said on May 4. Per PYMNTS and Reinhart, the experience assigns an agent or a team of agents to each shopper, ingests an event feed from across platforms and channels, and uses reinforcement learning to personalize the on-site browsing experience in real time.

Reinhart told PYMNTS that the system uses clickstream signals to predict "the path most likely to lead to conversion, changing what the customer sees as they navigate the site in real time." PYMNTS reports the agentic capability supplements ThredUp's existing AI tooling, which the outlet lists as product search, image search, ad buying, recommendations, photography, measurement and flaw detection.

Editorial analysis: Agentic experiences combine continuous event streams, online decisioning and reinforcement learning to adapt interfaces mid-session. Companies pursuing comparable real-time personalization often face engineering tradeoffs among latency, online feature stores, and robust experimentation frameworks to validate live interventions.

For observers and practitioners, key indicators will be measured conversion lifts from live personalization, on-site latency under load, how clickstream privacy and consent are handled, and the telemetry ThredUp or independent analysts publish on rollout scale and results.