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.
Institutions ask for a dashboard and discover they have bought a question. Here is the order of work that actually produces a number somebody is willing to sign.
The request usually arrives fully formed. Somebody has seen a dashboard elsewhere and wants one, with these charts, by the end of the quarter. The charts are rarely the hard part.
Acquisition first: finding every system that holds a piece of the answer, including the ones that are a spreadsheet on a specific laptop. Then cleaning, which is where the schedule usually slips, because the data is not wrong so much as inconsistent in ways nobody documented. Then modelling, which forces the organisation to decide what its own terms mean.
That third step is where the politics live. Two departments reporting different totals for the same quantity are not making an arithmetic error. They are using two definitions, both defensible, neither written down. A data platform cannot resolve that; it can only make it visible and insist somebody decide.
When a figure is challenged in a board meeting, the useful response is not confidence. It is the ability to open the number, see the transformation that produced it, and reach the source record in one step. Build that first and the reporting layer becomes almost trivial. Skip it and every disputed figure becomes a two-week investigation.
Ownership, classification, retention and access control are cheap to design in and expensive to retrofit. More to the point, retrofitted governance tends to be ignored, because by then everybody has working habits built around its absence.
Once the foundation exists, the reporting layer should be small. Three to five questions that genuinely drive the institution, answered clearly, opened daily. A dashboard with forty charts is a sign that nobody decided what mattered, and it will be abandoned within a quarter.
An imported crop classification model reported 94 per cent accuracy and was wrong about a third of the fields we checked. The gap was not the model. It was the ground truth.
Most exploration licences in the region sit on more geophysical and satellite coverage than anybody has looked at carefully. Fusing what exists is usually cheaper than acquiring more.
Digitization projects fail at indexing, not at scanning. A searchable archive requires decisions about classification that only the institution can make.
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.
