Nothing crosses the line you draw.
Your brand, your network
A white-label mobile or web app under your name, reached over your private network or by QR code.
Your knowledge bank wired in
HR, policies, products, services, and internal documents. Answers come from your material, with the source attached.
Models inside the boundary
Self-hosted or open weight, running on your infrastructure. Every instance is fully isolated.
Role-based access, and you keep it
Staff see what their role allows. The tools you build stay yours after the contract ends.
Price your private AI
Set your team size, usage, and knowledge depth. The bill recalculates at public list rates.
Write the rules once, enforce them everywhere: what the assistant may answer, per role and department.
Answers cite your knowledge bank. Off-limits topics are blocked at the boundary, not by policy memo.
Every conversation is logged and auditable inside your network.
Rules and knowledge update centrally. Every staff app follows instantly.
Frontier models stay stronger on hard reasoning. Most staff queries are retrieval, drafting, and lookup, where open-weight models hold up.
Open weights can run self-hosted inside your boundary: flat infrastructure cost, zero per-token licensing, no data leaving your network. Assumes 0.8K output tokens per request and 22 workdays. July 2026 list rates.
Where this sits.
Start with one team.
Tell us the department, the document set, and where it has to run. We come back with a pilot scope and a number.