Brett Bejcek

Brett Bejcek co-founded Limitless and led its user research and data science. His 2026 Techsylvania biography describes that work as taking the AI wearable from zero users to product-market fit. In 2026, he runs Click Clack Labs, a one-person Brooklyn studio spanning research, data, design, and code, and advises founders building consumer products. His professional focus includes choosing what to build, finding signals in small samples, and distinguishing meaningful retention from misleadingly encouraging curves.[3]

In his June 2026 update, Bejcek reported giving a talk on fulfillment in the age of AI. He found that effort did not explain why some AI-heavy projects left him proud and others hollow: “What separates them is whose decisions drove the work.” He also described an uncomfortable discovery about his own use of automation: “automating the parts I love and keeping the grind.” He expressed interest in creating guidelines for thoughtful interaction with AI.[4]

Before his solo studio work, Bejcek demonstrated Rewind at an AI×UX NYC meetup recounted by Paul Butler in June 2023. Rewind periodically captured screen snapshots, extracted text and visual context, and supplied that context to GPT-4. Butler presented the demonstration as an answer to a problem in personal search: information scattered across applications and drives, connected by context that conventional search indexes struggle to capture.[1][3]

Limitless was acquired by Meta in 2025. Its acquisition announcement also announced the sunsetting of Rewind; an update disabled screen and audio capture starting December 19, 2025.[3][6]

An earlier example of Bejcek’s machine-learning work was a collaborative summer project described in his August 17, 2017 account. The team used Seattle intersection sensors to identify vehicle cruising, including drivers searching for parking and for-hire vehicles operating without passengers. They reconstructed likely paths with a modified Dijkstra algorithm, manually segmented paths by time and transportation method to create ground truth, added trip metadata, and applied semi-supervised learning to label the data. A RethinkDB pipeline fed a Python Flask application with an interactive heat map showing relative cruising levels in downtown Seattle.[5]