Runs in your infrastructure
A dedicated deployment for you, in your cloud or on a virtual machine on your premises, or hosted by us. Chat history stays in it.
Private AI workspace · Saudi Arabia
A private AI workspace for organisations in Saudi Arabia
A workspace your security team can review: where it runs, who looks after it, and what the model provider receives. On the models you approve.
About an hour · Free · No installation
For IT and security
Where each part runs
Start here: where each part runs, who looks after it, and what the model provider receives.
In your deployment
Mazun Chat
Chat history is kept here
Mazun Gateway
Masks chat text, applies your rules
Masked chat text
Your approved model provider
Receives masked chat text only
Document checks: Word, Excel, PowerPoint and PDF files are checked before they reach a model.
Coming nextStarts in the UAE, on Mazun infrastructure.
Masking of supported data
Supported sensitive data in chat text is masked before a model sees it, including national ID and Iqama patterns and IBANs.
Blocked if a check fails
If a required check can’t run, the request is blocked and does not reach the model.
Your rules, applied
You set which files can be shared and the usage limits; private keys and common API keys are blocked too.
Model governance
Only the models you permit, from a curated list across vendors. Provider keys never reach staff.
AI insights
Mazun Cockpit reports adoption, usage and protection and exports to CSV. Admins sign in with multi-factor authentication.
Microsoft Entra or Google single sign-onPilot-ready
Mazun doesn’t see your prompts, conversations or files unless you give permission.
Built to support the confidentiality you owe your clients.
Your security review
What a security review asks, and what we can answer
Is a security review part of how your organisation adopts a new tool? These are the questions a review can ask, with the answers we can stand behind. We describe how Mazun works; we make no promise about regulation or data residency.
Where does it run?
In a dedicated deployment for you: in your cloud, on a virtual machine on your premises, or hosted by us. You choose where.
Who operates it?
Our team installs it, sets it up for your work and keeps it current every month. The operating scope, including response times and support hours, is agreed with you in writing; we do not state one here.
What does the model provider receive?
Masked text only. In supported chat text, Mazun Gateway masks supported sensitive data before a request reaches the model, and names come back in the answer so staff get a usable reply.
What happens if a check cannot run?
The request is blocked and does not go to the model.
What does masking not do?
It works on the formats Mazun recognises, so a detail in an unusual form can be missed. It reduces exposure; it does not remove every risk, and it is not a certification.
What you get
One workspace, set up and run for you
Mazun ChatModel: from your approved list
Summarise a contractDraft a board paperPrepare a tender response
Summarise this supplier agreement and list the obligations that fall on us.
Here is a summary of the supplier agreement. Four obligations fall on us: monthly reporting, a named contact, an audit window and notice of changes…
Ask anything
Mazun CockpitThis month
205Active staff
15,310Requests
1,104Details masked
11Requests stopped
Usage by team
LegalFinanceOperationsHRProcurement
Adoption lagging in Procurement: 4 of 22 staff active.
Illustrative screens and figures.
Mazun Chat
One familiar AI workspace for your whole organisation, on the models you approve, with company guidance built in.
Mazun Gateway
Sits between your staff and any model. Masks supported sensitive data in chat text and applies your policies first.
Mazun Cockpit
One clear view for whoever is accountable for AI: who uses it, how often, and where adoption is lagging.
Our team
We install it, configure it for your organisation and keep it current every month.
These aren’t four products. This is Mazun.
How to start
Start with one team.
About an hour
Free AI exposure check
We look at how AI is used in your organisation today and show Mazun working live, then send a short written summary.
30 days
30-day pilot
One team and one workflow, installed and supported by us, with the value measured before and after.
Every month after
Roll out
The rest of the organisation, with our team looking after it and keeping it current every month.
The business case is built from your numbers, not ours.
Cost
What AI costs your organisation now, against Mazun.
Filled after the exposure check
Risk
Which tools are in use, what client data goes into them, and who is accountable for the risk.
Filled after the exposure check
Value
One real workflow, with its time and quality measured before and after the pilot.
Filled at the end of the pilot
Questions
Questions from Saudi IT leads
Where does Mazun run, and who runs it?
In a dedicated deployment for each customer: your cloud, a virtual machine on your premises, or hosted by us. Our team installs, configures and keeps it current. For supported chat text, your approved model provider receives masked text only. Chat history is kept in your deployment.
What evidence can we give our security team?
A plain description of where each part runs, what the model provider receives, and what happens when a check cannot run. Book a meeting and we will walk your reviewers through it live. We do not offer a compliance certificate.
Does masking cover everything?
No. In supported chat text, Mazun Gateway masks supported sensitive data. A detail in an unusual form can be missed, and some text that is not sensitive can be masked. Images and voice are not inspected. Masking reduces exposure; it does not remove every risk.
Which AI models can our staff use?
Only the ones you permit, from a curated list across vendors and re-tested every month. Staff use the right model for the task. Provider keys never reach staff.
Can Mazun see what our staff type?
Not without your permission. If you ask for help that needs access, it is agreed in writing, limited to that purpose, those people and that period, and you can withdraw it at any time.
How does a pilot start?
A free AI exposure check of about an hour with someone from IT and someone from the business, then a 30-day pilot with one team and one workflow. You decide on a roll-out from the numbers.
Something else?hello@mazun.aiWhatsApp +971 55 524 6766
Your people have said yes to AI. Now your organisation can too.
Book a meeting
A free AI exposure check: about an hour with someone from IT and someone from the business. We look at how your people use AI today and show Mazun working live. You then get a short written summary: the cost and risk parts of your business case.
- Free, with no installation
- Mazun shown working live
- A short written summary afterwards
Prefer to talk?
Thanks. Your request is in.
We’ll reply within one working day from hello@mazun.ai to agree a time.
While you wait: message us on WhatsApp if it’s urgent.