A private AI workspace for universities and research institutes

Between student records and unpublished research, a campus holds two kinds of data that a public chatbot would happily copy off site. A name attached to a grade, a disciplinary note, a draft paper that has not cleared peer review yet: all of it carries weight, and all of it walks out the door the moment someone pastes it into a free tool to save twenty minutes.

You do not have to choose between that risk and falling behind. There is a way to give your staff and researchers a capable AI without sending a single sensitive line to a stranger's cloud. Keep the work, keep the data, and read on for how.

Why a public cloud chatbot clashes with a campus

A public cloud chatbot was built for one thing: send the text to a US data centre, get an answer back. That trade is fine for a recipe. It is not fine for a transcript, an admissions file or a grant proposal that names unpublished methods. Once that text leaves the building, you can no longer say where it sits or who else might train on it, and on a campus that holds both personal student records and original research, that is two exposures at once.

Banning the tools does not solve it. Your faculty, your administrators and your graduate students already use AI, on their own accounts, on their own phones, whether or not policy allows it. A ban just pushes that traffic out of sight, where you cannot govern it at all. The honest move is to give them something better that you actually control.

Run the model in-house

kral runs on your own server. The platform installs on your hardware, and you can add a local model on a machine you own, so a prompt about a named student or a line of unpublished research goes to your box and stops there. No external API sits in the path. Nothing is sent to a vendor to be logged or trained on. For the questions that must never leave, the request never does.

Most institutions mix the two. A cloud model handles the general work, drafting, brainstorming, formatting, where the input is harmless, and a local model handles the sensitive cases, where the input is a real person or unpublished result. You decide which model handles what, per team and per use.

A full workspace, not a chat box

kral is a full workspace, not a single chat box. Your team can build their own assistants in minutes, with no code: an assistant that drafts administrative communications so the registrar's office stops writing the same letter from scratch, or an assistant that summarizes internal documents you provide so a department head can read the gist of a long report in a minute. Save those as reusable routines and nobody on staff rebuilds the same setup twice. Drop in a document and ask questions about it. Pull a current, cited answer from the web when you need a fact checked. Switch between the leading models in one click when one handles a task better than another.

Connect your own systems

Your AI works best when it can reach your own material. kral supports MCP, the open standard for connecting tools and data to an AI, so an assistant can work with your own templates and internal knowledge through a connector you control, instead of guessing from the open web. The connection runs on your terms, and your systems stay yours. Nothing is copied out to make it work.

You run it and you see everything

You run kral, and you see everything in it. From the admin side you manage who is in and which models each person may use, set a spending limit per person so no single account runs up a surprise bill, and watch real usage on a dashboard. Single sign-on means staff use the login your IT already runs. It installs on Windows Server behind IIS, the same way your other applications already run, and it sits inside your network behind your firewall, in your own branding. If you want the wider picture for a campus IT team, here is the case for company-wide AI you host yourself.

We help you put it in place

You do not have to stand this up alone. We set kral up with you, connect it to your systems through MCP, and advise on rolling AI out across departments without the data leaving your side. Implementation consulting is part of what we offer, so the project lands instead of stalling in a pilot.

A capable AI and confidential data are not opposites. Run the model on your own server, keep student records and research where they belong, and give your people a tool they are allowed to use. Book a short demo and see kral on your own hardware.

Book a demo

Open the app