People stay accountable.
AI can propose. A named person decides.
About msharti
AI governance built for how Africa works.
Built with an understanding of local and regulatory context in Kenya and across Africa.
Book an assessment
Amina + AIReconcile M-PesamshartiAccess · approval · audit
SAPOnly permitted tools
A named person asks for one reconciliation. The agent requests a specific tool.
Why we exist
Built elsewhere is not built for here.
Most AI security tools are built for other markets, other systems and other rules. They arrive here knowing little about how organisations in Kenya and across Africa actually operate.
Meanwhile, AI is already at work in our offices, much of it never approved. Customer records, contracts and figures go into tools nobody chose and nobody can see.
We build AI governance and security for how organisations here work and are regulated. Our approach is informed by local law, such as Kenya’s data protection law, and by the expectations of financial-sector supervisors, so control fits the context instead of fighting it.
What we believe
Five principles we build by.
They shape every connector, every rule and every conversation.
Read first. Write with approval.
Access starts narrow and grows only when someone grants it.
Every action leaves a record.
If it cannot be explained later, it should not happen now.
Context comes first.
Local systems, local data, local law and local ways of working.
Honest about what is live.
We say what works today and what we are still building.
What we do
Two products. One promise.
AI with control.
AI Access Layer
Who and what AI can reach, with human approval and a full audit trail.
ExploreShadow AI Discovery
Finds the AI in use that nobody approved, and what it takes with it.
ExploreCustom Connectors
Your own systems, in plain language, under the same rules.
ExploreHow we work
Alongside you, one stage at a time.
You approve each stage before the next begins.
- 01
Assess
We find the AI already in use and what it touches.
- 02
Pilot
One team, one use case, read-only first.
- 03
Build
Roles, approvals and records set to your rules.
- 04
Run
We operate, monitor and improve it with you.
Who you work with
Close to the work. Close to the context.
The people who assess your environment are the people who build and run it.
One team, end to end.
No hand-offs between assessment, build and support.
Local by default.
We know the systems, the language and the rules you work under.
Work with us.
Engineers, security people and partners who care about getting this right.
Get in touchWhere we work
Kenya first. Then the region.
We start where we know the systems and the people, and grow from there.
NowKenya
NextEast Africa
LaterAfrica
See what AI already knows about your organisation.
A free Shadow-AI Exposure Assessment is the first step.