Software that widens the reach
and lightens the work.

We are an Enterprise Digitization Agency based in Nairobi, Kenya. We build the applications organizations use to reach more people, and we put agentic AI to work on the operations that quietly consume their week.

The problem

Most organizations do not have a technology problem. They have a set of processes that were designed for paper, carried into spreadsheets, and never revisited — and a service their people cannot reach because it only exists in one building.

Digitization projects stall in the same place every time: somewhere between a pilot that demonstrated well and a system nobody uses. We work on the second half of that sentence.

What we believe

Software nobody uses is a cost.Adoption is the only measure that matters. A system that is technically correct and quietly bypassed has made the organization slower, not faster.
Agreement beats assumption.We turn the diagnosis into acceptance tests and have you sign them off before we build. If we have misunderstood your problem, that is the cheapest possible moment to find out.
Automate the workflow, not the task.Agentic AI earns its keep on the whole path a piece of work travels — intake, checks, escalation, follow-up — not on one step in the middle that was never the bottleneck.
The system has to outlive the engagement.We hand over applications your team can run, extend, and argue with. We measure ourselves on what still works a year after we leave.

01 — How we work

Diagnose.

We map how the work actually moves today — who touches it, where it waits, and what it costs to wait. Most of the value of a digitization programme is decided here, before a single technology is chosen.

02 — How we work

Decide.

We make the call and put our name on it: what to build, what to buy, what to leave manual, and which workflows are genuinely worth handing to an agent.

Then we write it down as acceptance tests. Before a line of code is committed you sign off on the cases the finished system has to pass — so "does this solve the problem?" is a question we answer together, up front, instead of arguing about at handover.

03 — How we work

Apply.

Applications go into production, not into a slide deck. We build against the tests you signed off, and stay through integration, evaluation and the second and third iteration — the unglamorous part where most programmes quietly fail.

04 — How we work

Transfer.

The engagement ends when your team can run it without us. Documentation, handover, and the reasoning behind every decision left intact.

What we do

Four things we
build well.

Narrow on purpose. Each of these is a practice we run end to end — design, build, deploy and hand over — rather than a service line we can staff.

Applications

Custom Applications

Web and mobile products built for the people who will actually use them — on the devices and connections they have, not the ones we wish they had.

Automation

Agentic Workflows

AI agents put to work on the paths your operations already follow: intake, review, escalation and follow-up, with a human at the decisions that need one.

Digitization

Records & Process

Paper and spreadsheet processes turned into systems of record that are searchable, auditable, and available to everyone who needs them.

Platform

Integration & Data

The connective work: making the systems you already paid for talk to each other, and turning what they hold into something you can act on.

“Most of the organisations we work with already know what is broken. What they have been sold is a pilot, a licence, and a slide that says transformation. We would rather write down what the system has to do, agree on it with you before we build, and then be judged on whether the people doing the work can actually do it faster a year later. That is a harder promise to make, and it is the only one worth making.”

Breldan Muturi Muthaka · Founder, Sohn & Sol

Selected work

Built, not advised.

Three engagements where the work went into production and stayed there — a product suite for a power generator, an accessibility remediation that opened up employment, and AI put to work on physical security.

Energy · KenGen

A product suite, not a website

Book-a-tour and the geothermal spa opened Olkaria to the public. The careers portal put AI behind recruitment, screening and routing applications for a workforce of 2,500 across decentralised stations. Each product reached an audience the corporate site alone never could.

Public sector · NCPWD

Accessibility, then employment

We remediated every accessibility violation on the National Council for Persons with Disabilities website, then integrated it with Fuzu — so the people the Council exists to serve can find and apply for jobs on a platform that no longer shuts them out.

Banking · Consolidated Bank

AI watching the premises

AI-enabled CCTV across the bank's premises: video analytics that flag what matters in real time instead of leaving hours of footage to be reviewed after the fact — the same agentic thinking we apply to workflows, pointed at physical security.

Research & articles

Things the industry
keeps getting wrong.

01.

Your bottleneck is not the step you automated.

Teams reach for AI at the most visible task and are surprised when the queue does not move. The delay usually lives in the wait between steps, not the steps themselves — and that is a workflow problem an agent can solve, if you point it at the right thing.

Read article
02.

Build for the connection your users actually have.

Reach is decided by what an application demands: page weight, offline behaviour, the assumption of a recent device. In markets where the median user is on intermittent mobile data, these are not optimizations — they are the difference between a service that exists and one that does not.

Read article
03.

The pilot was never the hard part.

Demonstrating that something works is cheap. Making it survive real volume, staff turnover and the process it replaced is the work — and it is the part most digitization programmes are not scoped to finish.

Read article