
Hedy runs its EU AI processing on Inceptron
We’re partnering with Hedy, the AI meeting coach, to run cloud AI processing for its EU data region.
Hedy listens to meetings on the user’s own device, transcribes them locally, and uses AI to suggest questions and talking points during the conversation. After the meeting, it generates summaries and notes.
For accounts in Hedy’s EU data region, that AI processing now runs on Inceptron.
What’s live
Hedy is moving its EU traffic to Inceptron in stages.
Its EU region currently generates approximately 6 billion tokens per month, with usage continuing to grow.
Why Hedy chose Inceptron
Hedy is a US company with a significant share of its users in Europe.
As part of its search for an EU inference partner, one of Hedy’s main requirements was straightforward: EU-only processing had to be covered contractually for the serverless API it uses.
Our DPA with Hedy specifies that its account is configured for EU-only processing.
“Our European users tell us where their conversations are allowed to go, and for most of them the answer is the EU. We wanted that written into the contract for the exact service we use, and Inceptron was ready to do it.
Their prompt caching also suits how Hedy works, since a live transcript keeps growing through a meeting.”
Julian Pscheid
Founder & CEO, Hedy AI
EU-hosted inference
Processing for Hedy’s account stays within the EU/EEA.
Request and response payloads are not stored after processing by default, and customer content is not used to train models.
For Hedy, this provides a clear deployment model for European accounts without requiring separate infrastructure or a dedicated cluster.
Built for long, growing prompts
Hedy’s workload is also a natural fit for prompt caching.
During a meeting, the transcript grows with every request. Much of the previous context is reused, which makes prefix caching particularly effective.
In Hedy’s own testing, approximately 90% of input tokens were served from cache from the second request onward.
That reduces the cost of repeatedly processing the same conversation history as the meeting progresses.
Read more
Hedy has also published its own announcement about the partnership: