A private AI workspace for local government

Citizens have to give you their data, they did not agree to it being copied onto a private chatbot company cloud. They hand over an address, a health note, a benefits claim, a complaint about a neighbour, because the law and daily life leave them no other door. They trust that it stays inside the building.

The moment that text is pasted into a public chatbot, it leaves. It lands on a server you do not control, in a country you did not pick, under terms nobody at the council signed. There is a way to give your teams a capable AI and still keep every word about a resident on your own side. It comes down to where the model runs.

Why a public cloud chatbot clashes with citizen data

A public AI chatbot ships every prompt to a vendor's cloud, usually in the US. For a private company that is a vague worry. For a council it works against the basic promise you make to the people you serve: that what they tell you stays with you. You cannot say a case file is held in confidence and then route a question about it through a service you have no contract with and no view into.

Banning AI does not solve it. Caseworkers and clerks already use it on their phones to word a tricky letter or make sense of a long document. The honest question is not whether your people use AI, but whose machine it runs on.

Run the model in-house

With kral, the whole platform runs on your own server. You can add a local model on your own hardware, so a prompt about a named resident goes to your machine and stops there. No external API sits in the path, nothing about that person leaves the building. Most teams mix the two: a strong cloud model for general work where no personal detail is involved, and a local model for the sensitive cases tied to a real name.

A full workspace, not a chat box

This is more than a box to type into. Your team can build their own assistants in minutes with no code. One assistant drafts citizen responses in plain language, so a benefits decision or a planning reply reads clearly instead of in jargon. Another summarizes a policy document for staff, turning forty pages into the points that matter before a meeting. Save those setups as reusable routines, so nobody rebuilds the same thing twice.

Drop a document in and ask questions about it. Pull a current, cited answer from the web when a query needs one. Switch between the leading models in one click when one handles a task better than another. Everything sits in one place, behind your own login.

Connect your own systems

kral supports MCP, the open standard for connecting tools and data to an AI. The assistant can work with your own templates and internal knowledge through a connector you switch on and control, instead of guessing from the open web. Your letter formats, your service records, your way of doing things feed the answer. Your systems stay yours.

You run it, and you see everything

You manage who is in and which models each person may use. Set a spending limit per person so costs stay predictable, and watch real usage on a dashboard instead of guessing. Everyone signs in once through single sign-on. It installs on Windows Server behind IIS, the same approach as a company-wide AI you host yourself, and sits inside your network, behind your firewall, in your own branding.

We help you put it in place

You do not have to work this out alone. We set kral up with you, connect it to your systems, and advise on rolling AI out across the team without the data leaving your side. Implementation consulting is part of what we offer, so the first local model and the first connector are running before your people ever touch it.

Give your teams a capable AI workspace and keep citizen confidentiality intact. See it running, then let us help you put it on your own server.

Book a demo

Open the app