The enterprise OSgoverned by your data

How do we align teams in an AI world?

Give people and AI one governed workspace: shared projects, chat, docs, sheets, slides and mail, grounded in your own data from 17 database engines, on your own servers.

No AI model lock-inSovereign: runs on your serversEvery AI request governed

Project · Q1 board packTNJSLB
One project, one set of rules
TNThandi
@OQLIS why did gross margin fall in March?
Freight rose 19% under the new carrier contract, mostly on the Walvis Bay route. Prices didn't change. Certified metric · Gross margin
JSJohan
Put that on slide 4 and update the letter to the board.
3 changes suggested in the slides and the letter. Accept or reject each one.
Review changesOpen slides

Three questions every business asks about AI

AI can draft, answer and act. The hard part is trusting it, connecting it to what you already run, and letting people use it without losing control.

1

Can we trust what AI produces?

Answers about your business come from certified metrics and the rows behind them, shown as a live chart or table. The assistant is told never to invent a figure, and its edits to a document arrive as suggestions a person accepts.

Answers from your numbers

2

Can we use the systems we already run?

Yes, without replacing them. OQLIS reads 17 database engines in place, works in Gmail and Outlook, Google Drive and WhatsApp, and turns your own scripts and tool servers into tools the assistant can use.

Everything it connects to

3

Can our people work safely with AI?

Every message runs as the person who sent it, with their access and their budget. Mail is never sent and systems are never changed without a person, and every AI request is on record.

AI you can account for

An operating system for the business, built on its data

Most data tools stop at the dashboard. In OQLIS, the harness does the work that follows: it plans, waits for a person where it matters and builds the result from your certified numbers.

What we mean by an enterprise OS

The harnessEvery task OQLIS runs for you, in one place

Each task shows what it read, the agents working on it and what it wants to do next. Anything that would change your systems waits for a person.

See how agents work

Hand over the long work. Agents split it up and report back when it is done.

Shows the plan and cost first

Before long work you see the steps and an estimated cost, and you decide whether it starts.

OQLIS
Create
Today, 10:20
Compare Q1 margins by region and draft a page for the board.
Plan: 2 agents in parallel, about 8 minutes, estimated cost $0.65. Shall I start? finance.marginsCost reports
Yes, start, and share the page with finance when it is ready.
Ask OQLIS

Splits the work across agents

Agents work in parallel and share notes, and keep going if you close the tab.

Lets you try a task first

Test a new task without it sending or changing anything, then put it live when it is right.

Supplier price watch
3 of 4 on
  • Try it firstSends and changes nothing
  • LivePuts the task to work
  • Start on a price list changeChecked before any AI is called
  • Weekdays at 9:00Schedule

Tasks report only what matters

Tasks stay quiet unless there is a result or one question for you.

  • One question from a taskAdd the two late orders to the digest?
  • Weekly sales digest is readyA page in Docs, shared with Sales

Your AI, your rulesAny model, on your servers, under your governance

Every AI request in OQLIS, from chat, agents, flows or WhatsApp, passes through one gateway that you own. That one choke point is what keeps you free to choose, sovereign over your data, and in control.

  • No model lock-inAnthropic, OpenAI, Google or xAI with your own keys, or open models on your own hardware. Policy maps each application to its models, with a fallback, so changing provider is one change in one place.
  • AI sovereigntyOQLIS runs inside your network, even with no internet. Your keys stay on your server, and with local models no prompt ever leaves it.
  • Complete governanceEach request is checked for role and budget, stopped if it carries sensitive data, and logged with its cost. An AI register keeps you ready for the EU AI Act and ISO/IEC 42001.

See how Control governs AI

The data platformGrounded in one set of numbers, from the databases you already run

Under the harness is the data platform customers have bought OQLIS for over the years. It reads your data where it lives, your data team agrees each measure once, and every answer, dashboard and alert uses it.

See the data platform

1 · Connect

Connections, read where they are

17 database engines, queried in place. Nothing has to be copied into OQLIS first.

2 · Model

Certified metrics, agreed once

Certified metrics with an owner, so finance and sales quote the same figure.

3 · Use

The same figure everywhere

The same figure on a dashboard, in an answer and in an alert, filtered to each person.

For every roleEveryone gets the view their job needs

One product with the same data and rules, shaped for each person's work. Each workspace decides which roles use which.

Explore each perspective in depth

Read dashboards and reports

See what your team has published, and ask questions about it.

  • Dashboards and apps
  • Ask about any chart
  • Share with a read-only link

Who uses it: Everyone

Explore View

Where it runsYour data, your servers, your name

OQLIS began as a data platform, and that is still what customers buy it for: their own numbers, read where they live, on their own servers and under their own name.

"We've tried all the big players and they couldn't answer our questions. As soon as we met with OQLIS we realised these guys could answer our questions."
Melvin April
Namibia Institute of Pathology, on its own servers since its proof of concept

NAD 2M+

a year in licence fees replaced

What customers got from it

From pathology labs to security operations centres.

All customer stories
  • Namibia Institute of Pathology
  • Datacentrix
  • Zutari
  • Unitrans
  • Forte Supply Chain Solutions
  • G4S
"The biggest differentiator for us with OQLIS was the ability to give our customers the opportunity to play with their own data, without breaking anything."
Kyle Pillay
SOC Manager, Datacentrix

Questions buyers ask first

Something else on your mind? Email us a question and the team that builds OQLIS will answer.

Do we have to move our data?

No. OQLIS reads your databases in place, with a read-only user.

What does OQLIS connect to?

17 database engines, from SQL Server and SAP HANA to Snowflake, plus mail, calendars, Drive, WhatsApp and MCP servers.

Can we put our own brand on it?

Yes: your name, logo, colours and web address.

Where does OQLIS run?

On your own Linux server, even offline, or in OQLIS Cloud.

How is OQLIS licensed?

With unlimited users: no charge per person.

How soon can we see it on our data?

Most customers start with a short proof of concept. The Namibia Institute of Pathology took three weeks.

Can the assistant change things alone?

Not by default. Changes wait for a person, and each one is logged.

Are we locked into one AI provider?

No. Use Anthropic, OpenAI, Gemini or xAI with your own keys, or open models on your own hardware, and switch in one place.

Can we use it away from the browser?

Yes: on WhatsApp, and in the Mac and Windows apps.