Tardi

An AI that fits in a phone and works offline.

Tardi understands the language of documents, manuals and records to find, compare and decide. It runs on the CPU of a laptop, a server or a phone, so information never leaves the organization's network.

Offline
No GPU
On-device
100% private
Spanish and English

Book a briefing Go to tardicompute.com

Tardi · AI labtardicompute.com

Model size, to scale

Open frontier model~1 TB
Mid-size local model~141 GB
Small local model~16 GB
tardi core15 MB
tardi nano6.9 MB

tardi nano is about 150,000 times smaller than an open frontier model. That is why it runs on a phone, offline.

Built with Model Foundry

Model FoundryTardi comes out of Model Foundry, the factory that rebuilds advanced models around a single problem.Discover Model Foundry

Let's look at your case with your own data.

In a briefing we review with your team which sources you have, which questions you need answered and what we can deploy.

Book a briefing contacto@naboointel.ai