← All organizations

AI software testing and developer tools

Tusk

Tusk built AI agents that helped engineering teams generate and run unit and integration tests within their pull-request workflows. Its agent used codebase context and business logic to suggest missing tests, while CoverBot identified under-tested files and opened pull requests to backfill coverage. Tusk Drift recorded API traffic and exposed it through an MCP server, letting coding agents build test fixtures from actual requests, responses and database queries. Tusk announced it was sunsetting its product effective June 10, 2026, ending its availability as a standalone testing service.

Founded in 2023 by Marcel Tan, its CEO, and Sohil Kshirsagar, its CTO, Tusk participated in Y Combinator’s Winter 2024 batch. It initially automated engineering tickets into pull requests before focusing on testing. Its technical approach combined execution with production context: generated tests ran in ephemeral sandboxes and were revised iteratively, while Drift supplied real integration data to reduce incorrect mock values. Recorded traffic complemented code analysis but could miss failure scenarios absent from captured requests.

In a case study published in 2026, Tusk reported that customer Hamming added 1,800+ tests over three months using CoverBot, with a 66% merge rate for its pull requests. Y Combinator lists Tusk as acquired; its profile does not identify the buyer or transaction price.

www.usetusk.ai

1 talk

Newest first

1 speaker at AIE

Affiliations reflect their AIE appearances, not necessarily current employment.

Messages from the stage

Prioritizing feedback and finding edge cases

Tan described verified AI-generated tests, systematic pessimism for discovering edge cases, and a novelty-criticality framework for deciding which feedback is useful.

Affiliations reflect each recorded session, not necessarily current employment.

Company sources · checked 2026-08-28