Varun Shenoy is a Long Lake cofounder working on the adoption of AI inside services businesses. His earlier work spans medical imaging and model infrastructure.[2]
Start with How do you diffuse AI into the real world? →
X · varunshenoy_ Career
Before Long Lake, Shenoy built medical-imaging tools at the U.S. Department of Veterans Affairs and worked at Baseten, Doss and Berkeley AI Research. He studied electrical engineering at Stanford. Long Lake acquires and operates services businesses, placing the technology team alongside the people whose work it seeks to improve.[2][4]
Ideas
His conference talk distinguishes an impressive agent demonstration from changed daily work. He proposes progressing from immediate assistance to synchronous, asynchronous and longer-running tasks as the system earns trust. Operational traces supply explicit feedback and implicit evidence, such as the difference between a generated record and the final corrected record. These traces can support evaluations and model improvement. Employees also need help adopting useful workflows. The claim concerns organizational adoption alongside model capability.[4]
Tools, writing & talks
Writing, research & talks
Start here
How do you diffuse AI into the real world?ai.engineerA services-company account of increasing agent autonomy, collecting operational feedback and helping employees adopt new tools.
Start with the autonomy progression and the examples of operational traces. Basic agent familiarity helps; business operators can use the workflow questions without implementing the model-training layer.
Speaker[4]Long Lake buys the services business rather than selling it software, across 35 companies and a $6.3B take-private, so a failing agent has no customer to blame. Shenoy's argument: the valuable work — closing books with missing receipts, scoping a building from a blueprint — is absent from the intern
Catalog workOriginal source