Ten capability lines, grouped by what the systems do: reason, remember, observe, hold.
Akili Smart Solutions Limited. Registered in Tanzania, building applied intelligence for industry, government and research.
The sequence is short on ceremony and unforgiving about evidence. No stage begins until the one before it has produced something a client has read and signed.
The common way for a technology programme to fail here is not bad engineering. It is eighteen months of good engineering aimed at a decision nobody had written down.
So the first two stages cost very little and are allowed to end the engagement. That is unusual and it is the point: we would rather lose a project in week three than deliver a system in month eighteen that no one opens.
After that the work is ordinary, disciplined delivery, in increments you can see.
The right-hand column is the deliverable, not a description of effort.
Establish the decision before the technology.
We start with the decision that is currently going wrong, who makes it, on what evidence and how often. If that cannot be written in a paragraph, there is no brief yet and no amount of engineering will supply one.
Find out what the data can actually support.
An audit of the data that exists, its quality, its lineage and its gaps, run against the question from stage one. This is where most ambitious programmes are either rescued or quietly rescoped.
Design for the conditions, not the brochure.
System design against real constraints: bandwidth, power, device estate, existing systems, budget and the skills of the team who will inherit it. We cost it before anybody commits.
Short cycles, working software, visible progress.
Delivery in increments a client can see and use, with testing, documentation and security treated as part of the build rather than a phase at the end of it.
Leave the institution able to carry it.
Training, runbooks, source, data and the support arrangement. The measure of a finished engagement is that the client could continue without us, and chooses not to on the merits.
If any of them cannot be answered, the engagement pauses rather than proceeds. Pausing is cheap. The alternative is not.
Scope drifts quietly. We re-read the problem statement at every gate and amend it in writing, or we stop and re-frame.
Data that looked adequate in the survey sometimes is not. Better to find that at a gate than in production.
If the answer is no at stage four, it will still be no at stage five. Capability transfer is measured throughout, not at the end.
Sunk cost is not an argument. Either the rest of the programme pays for itself from here, or it should be cut here.
If yours is not here, ask it directly. We answer the awkward ones in writing.
Usually two to five weeks together, depending on how many systems hold the data and how reachable the people who use them are. It is deliberately short. If it starts to run long, that itself is a finding about the state of the data, and we say so rather than billing through it.
We tell you, in writing, with the measurements that led us there, and we stop. A short study that prevents a long failed programme is the most valuable thing we can sell. You keep the findings and the data inventory either way.
Routinely. Most institutions run a landscape they did not design and cannot replace in one step. Integration with what is already there is normal work, and we will run an implementation that includes vendors we did not choose.
You do. Source, data, models, documentation and infrastructure configuration transfer at the end of the engagement, and the agreement says so before it starts. Where we supply a licensed component we name it, its cost and its renewal terms up front.
Named people inside your organisation working alongside our engineers for the final stretch, written runbooks for every routine operation, recorded walkthroughs of the system internals, and a support agreement whose purpose is to become unnecessary.
Sometimes, and we will check the first two stages quickly rather than skip them. A brief that arrives fully formed is usually describing a solution, and the half hour it takes to confirm the underlying decision has saved several engagements from building the wrong thing well.
Not the technology you think you need. The decision, who makes it, and on what evidence. We will tell you plainly whether there is a system in it.
