Your data stays in-house
Runs entirely on your own hardware. Documents, knowledge, and agents stay put — with role-based access so the right people see the right things.
On-Prem AI Platform
The complete, on-premises AI platform for building purpose-built agents from your own business knowledge — for any team, any use-case. Everything out of the box, on hardware you own.
Each team builds its own agents — customer support, employee onboarding, product recommendations, an internal help desk, policy & compliance Q&A — each tuned to its own knowledge. Solve one problem, then the next. No new tool each time.
Runs entirely on your own hardware. Documents, knowledge, and agents stay put — with role-based access so the right people see the right things.
Turns your PDFs and scans, Word and Excel files, spreadsheets and images into agents that speak your business — in your terms, not a generic model's.
Grounded, source-cited answers, curated by your own experts so accuracy compounds — with built-in quality metrics you can measure, not just trust.
Beyond documents — Vesta reads your spreadsheets and tables and answers analytical questions: counts, trends, and breakdowns, worked out from your real data.
Tell Maya what you're trying to do and point her at your documents. She proposes a model of your domain and a first agent tuned to your business — you approve or refine, and it's answering real questions while you watch.
Put agents to work across your business
One product, sized to the hardware you have. Every edition uses the same document representation, so moving up a tier never means re-processing your library — and every edition is fully air-gapped once installed.
The whole platform on a single workstation you own, built from off-the-shelf parts. One box serves one organisation.
| GPU | NVIDIA, 16 GB VRAM — RTX 5070 Ti reference. Below 16 GB is not supported. |
|---|---|
| CPU | 6-core x86-64 minimum, 8-core recommended |
| Memory | 32 GB DDR5 minimum, 64 GB recommended |
| Storage | 1 TB NVMe SSD (2 TB recommended) |
| Software | Windows 11 + WSL 2 + Docker Desktop with GPU support, current NVIDIA driver. Linux with Docker is also supported. |
| Licensed to | your GPU's unique identifier |
Reference machine ≈ US$2,000 Approximate one-time hardware cost, from off-the-shelf parts. Vesta is licensed separately.
The same Vesta on Apple Silicon, with the models running natively on Metal rather than in a virtual machine.
| Machine | Apple Silicon Mac (M-series) — Mac Studio, Mac mini or MacBook Pro |
|---|---|
| Memory | 24–32 GB unified memory or more. The models, macOS and the container stack share one pool, so 16 GB is not enough. |
| Storage | ~50 GB free for the software and models, plus room for your documents |
| Software | Docker Desktop with Model Runner enabled, Xcode Command Line Tools |
| Licensed to | your Mac's hardware identifier |
Runs on a Mac you already own No dedicated GPU and no separate machine required.
Larger multi-GPU tiers, for bigger corpora and more concurrent users, are available separately. Tell us what hardware you have and we will tell you honestly what it will and will not do.
Your licence is a signed file. Vesta checks it on your own machine, with a key built into the product. There is no activation server to reach and nothing to phone home to.
A licence is issued against your machine's unique identifier — you send us a fingerprint once, and we send back a licence that runs on that hardware.
Every release is free while your term is valid, and updates preserve your data. If a term lapses there is a grace period, and your data is never touched.
The full detail is public: Administrator Guide. Internet is needed only to install — after that, nothing leaves the machine.
Three steps, and the first one takes a minute.
Run one small script and send us what it prints. It reads nothing but your hardware's identifier.
We send back a licence locked to that machine, along with download links for the release.
Download, unpack, run the installer. The commands are written out in full, with nothing to work out.