Sona8 uses voice agents to uncover how employees actually work

Sona8 is putting a voice agent in the employee interviewer's chair to help teams understand their business before changing or automating it. The startup, part of Y Combinator’s Fall 2026 batch, captures employees’ knowledge of everyday work and turns it into a clearer picture of what could be improved.

The problem is familiar to anyone who has worked on a transformation project. Official procedures often leave out the exceptions, informal decisions and workarounds that keep an organization running. Much of that knowledge stays with the people doing the work.

Customers choose the topic they want to investigate and invite relevant employees to a conversation. The AI conducts interviews in the participant’s language, asking for more detail when an answer needs explaining. It might explore why information is entered twice or how a request moves between departments. The resulting analysis helps teams identify recurring issues and compare workflows.

Screen sharing now lets employees demonstrate a task while discussing it. Camera support brings physical spaces into the conversation too. Teams can follow up during a rollout to understand whether changes are helping and what problems remain.

The founding idea came from Anton Hantel and Thilo Tamme’s work together at BCG. They had experienced how much effort went into gathering employee perspectives and keeping an accurate view of a transformation as it unfolded. They began exploring whether AI could make those conversations possible across an entire organization.

Hantel, the CEO, started building Sona8 alongside his MBA at MIT. Tamme, the CPO and a former Project Lead at BCG X, was researching conversational AI at the Technical University of Munich. Both left their programs as the pilots gained traction. CTO Madeleine Malmsten, a self-taught engineer and repeat founder, previously served as a principal engineer at QuantumBlack, McKinsey’s AI unit. The team combines direct experience of organizational transformation with conversational AI research and enterprise software development.

Their ambition reaches beyond individual interviews. Sona8 aims to build a company brain: a living understanding of the business that combines employee knowledge with data from its systems. Knowledge gathered during one project could then support the next, giving internal teams more capacity to do work that would otherwise require another consulting exercise.

That understanding would also serve as a context layer for AI agents. Before automating an approval process, for example, an agent would need to know which exceptions matter, why they exist and who relies on them. The goal is for different tools to draw on the same company knowledge, without repeatedly asking employees to explain their work from scratch.

The interviews and analysis are available today; the broader company brain is the direction Sona8 is building toward. Its usefulness will depend on whether that understanding leads to better changes in practice, and whether it stays accurate as the business evolves.

Learn more at sona8.com.

‍