AI & intelligent automation
We use AI where it saves real time: reading and sorting documents, drafting routine replies, and searching across your own records. We test each system on your real examples before it goes live, and important decisions are always reviewed by a person.
- Document-heavy processes
- Customer support
- Searching internal knowledge
- Back-office work
Before and after
The same documents and the same team, with the repetitive part handled by software and people kept in charge of decisions.
Before
- Documents typed in by hand, one at a time
- Files sorted and emailed around manually
- Answers depend on who remembers where things are
- Mistakes found weeks later, if at all
After
- Documents read and filed in minutes
- Each one sorted into the right place
- Answers found across your records, with the source shown
- A person approves anything important
What you get
- Finding the right use case
- Document reading and classification
- Assistants that answer from your own data
- Accuracy testing before launch
- Review and approval queues
- Reporting on usage, cost and accuracy
How we work
Tested on your data first
We agree an accuracy target with you and test against real examples before anything touches production.
A person signs off
High-stakes actions wait in a queue for approval, and every decision the system makes is logged so it can be reviewed later.
Careful with personal data
We minimise and redact what is sent to AI providers, in line with the Kenya Data Protection Act and your own policies.
Typical technology
- Claude
- OpenAI
- Python
- Vector search
- PostgreSQL
- Evaluation tooling
Have a project in mind?
Send us your brief, terms of reference or RFP. We reply within one business day.