Context layer for enterprise AI surveying employees with voice agents
TL;DR: Sona8 is building the context layer for AI automation tools, starting with AI voice surveys of employees. Corporate development, transformation and consulting teams use Sona8 to find out how to improve or automate processes.
https://www.youtube.com/watch?v=jGs7M46UsSE
We're Thilo, Madeleine and Anton. We're building Sona8.
At BCG and McKinsey we often interviewed a few dozen people, hired interpreters and came back with a deck. The processes that were written down were never the lived processes and a few dozen interviews never gave us the full picture.
How it works
- Choose what you want to improve or automate. Pick a question and the people to survey. This could be about sales quoting or mapping processes before implementing AI.
- Survey employees with AI voice agents. Employees talk to our voice agent about their work, in their own language. The agent asks follow-up questions to understand the problems they describe.
- Use the findings and check progress. See themes across interviews by department, get process maps, and ask again later to see what changed.
Already live
We’ve completed more than 10,000 employee interviews and signed one of the top three consulting firms.
Where this is going
Companies will try a lot of AI agents and automation tools in the next few years. Each one needs to know how the work is done before it changes anything, and most of that isn't written down. Sona8 creates infrastructure for AI automation tools. Over time we are building a live context graph of workflows by reading the systems a company already uses and connecting them to what employees say.
Who it’s for
Sona8 is for corporate development and transformation teams in enterprises, and the consulting firms supporting them. Reach out if you want to understand and improve how work gets done, roll out AI, or build AI tools that understand real-world workflows.
👉 Book a demo / anton@sona8.com
Team
- Anton Hantel (CEO): ex BCG, MIT MBA (dropped out)
- Thilo Tamme (CPO): ex BCG X, TUM Conversational AI PhD (dropped out)
- Madeleine Malmsten (CTO): ex McKinsey QuantumBlack
- Jakob Schepers (Founding Engineer): TUM CS Master's
- Niklas Schneider (Founding Account Executive): ITONICS, Maastricht SBE
How did your company get started? (i.e., How did the founders meet? How did you come up with the idea? How did you decide to be a founder?)
Thilo and I worked on the same consulting project. We did over 70 stakeholder interviews, then had to manage the stakeholders during implementation, which meant aligning the same people over and over. In the nights we went running a lot in that time. On one of those runs we started asking what if we automated the whole thing. An AI would be strictly better at every step. It asks better questions because it has heard the other interviews, it analyzes with all the context, and it can go back a month later and ask again.
What is your long-term vision? If you truly succeed, what will be different about the world?
Every big decision in a company is a choice between alternatives. The alternatives you see depend on the context you have. The context that matters most, how the work is done and where the problems are, sits with thousands of employees and never reaches the people who decide.
If we succeed, that context reaches them. Every employee talks to Sona8, and Sona8 plugs into the ERP, the HR system and the ticketing tools. Together that is a live map of how the company runs, and far more than any team can read. The hard part is finding, out of all of it, what matters for one question, and then adding the atomic insights and the judgment. This way, transformation, strategy and consulting teams operate far leaner and faster. Three people steer a program across 20,000 employees.
On top of the map we build a synthetic layer of employees. Before a decision, you run it against that layer and see what it does across sites and teams.
The same map is the enterprise context layer for AI agents. Before an agent changes a process, it reads how that process is done today and by whom. Sona8 is where that map lives.
