Deepgram targets air-gapped voice AI on laptops

Deepgram's push toward laptops comes after 13 years of tuning voice models for noisy, high-stakes settings such as air traffic control and emergency medical services. The company expands its Flux model to 10 languages earlier this year and assumes any model it ships will soon be surpassed, treating each release as a candidate for retraining.

That infrastructure need spans both its in-house research team and tens of thousands of external customers. A large percentage of those customers host the models themselves, and a recent partnership with Fortanix Inc. brings private, on-premises voice AI to regulated industries. Dell Technologies supplies Deepgram's infrastructure while also running its models in Dell's Customer Solution Center Innovation Lab, a reach that Mann described as ranging from enterprise to small companies.

The laptop goal sits at the far edge of that spectrum. Today, a small business can handle interactive voice response on a compact Nvidia GB10 system, but Mann said the larger problem he is chasing sits outside the data center, in the AI-enabled laptops students now carry to school. Running models natively and air-gapped there would be a different deployment path from the cloud, on-premises and Dell-hosted setups Deepgram currently supports.