Short pieces for the people who own an operation: what to automate first, what a voice agent can honestly do, what sovereign really means, and where the money goes.
What you will find here
Filter by topic. Each piece stands on its own.
Text-to-speech is excellent in about twenty languages and thin everywhere else. Some of the places where it is thin are rich, which is why custom voice training is a business. A map, with the caveats.
Most companies do not have an AI problem. They have a choosing problem. A short method for picking the first job, and the three tests it has to pass.
Real-time speech models changed the ceiling. They did not remove the floor. The honest list, for anyone about to put an agent on a phone line.
"Your data stays in the building" is a promise with five parts. What each one means, and the questions to ask any vendor who makes it.
The biggest model is rarely the right one for a job that runs ten thousand times a day. Why we distil, when we do not, and what you own at the end.
The interface that makes agentic AI work in a business is not a chat box. It is a queue with a person on it. What that looks like, and why it changes the outcome.
How we train, serve and ship. Inference, speech, post-training, evals and air-gapped deployment, with the code and the failure modes.
Engineering blogTwenty minutes is enough to say whether we can take it.
A pilot starts within 5 working days of agreed scope · Nothing upfront · No seat licences
