Roles are redrawn around human skill, not handed to agents as drop-in replacements. The work is redivided by what each does best. That is the horizontal cut.
In the risk functions we have worked inside, the work of assembling data consumes the part of the week that should belong to analysis. The work is beneath the people doing it — so the platform takes the assembling, and the function keeps the judgement.
Drawn from our own experience inside enterprise risk functions — an observation, not a statistic.
Most AI projects take a whole role, or a whole step, and hand it to a machine. We cut the other way. We go across the process and split each step by the kind of work in it. Judgement, relationships and accountability stay with people. Retrieval, drafting and calculation go to the machine. No role is handed over whole, and the person keeps the part of the work that needed them.
Because we start from the outcome rather than the existing steps, the same move does three things. The result gets better, not just faster. The grind moves to the machine. And people are left with work worth doing.
We apply this first between the vision at the top and the systems of record underneath: the strategy, planning and decision-making that today gets rebuilt by hand every cycle. Those systems stay. We fill the gap between them, where judgement matters most and the grind is thickest.
Our products are layers of one system, built in sequence: discover where you are, create the vision, model your processes, build the infrastructure, run it as a platform. It is designed so that what we learn running it flows back into the model. The grid below shows where each layer stands today.
Select a column to expand it.
