The diagnosis

Two weeks. One function. A decision you can act on.

Most AI programmes start with a tool and work backwards to the value. This starts with the work — read task by task — and tells you where AI belongs, what to change, and what it is worth.

What it answers

Three questions, answered at the level of the actual work.

01

Where should we use AI?

Not which tool to buy. Which steps of which workflows AI can carry today, which become a partnership, and which stay human — every task scored, so the answer holds when someone senior pulls on it.

02

What should we change?

The future-state design for the workflows that matter most: who decides what, where exceptions route, what the handoffs become, and which teams are still viable in their current shape.

03

Where will it actually create value?

Per workflow, in a range, with every assumption on the surface and upside kept out of the base case. Plus what has to be true before the value can be captured, and what should wait.

What lands on your desk

Not a report to file.

You get a small set of resolved decisions, each with a verdict, the named bets it rests on, and the working behind every number. Underneath them sit the blueprints the decisions came from.

What the two weeks produce
01The function blueprintThe future-state operating model, and a viability verdict for every team in the function — which are durable, which are time-bound, which need redesign, and the evidence behind each call.
02One or two workflow blueprintsYour most critical workflows, step by step: where AI leads, where people decide, how the handoffs change, cycle time today against cycle time after.
03Three to five role blueprintsThe roles that run those workflows, at task grain: what AI absorbs, what it augments, and which human capabilities become more valuable as it does.
04Per-workflow value, in rangesWhat each redesign is worth and what it costs. Conservative to optimistic, every assumption named, upside we cannot defend kept out of the base case.
05The decision readoutA leadership session working through the resolved decisions in capital-allocation order. Everyone reads the same evidence from where they sit.

Open the Meridian sample to read a finished one end to end.

What we need from you

Less than you would expect.

  • Your sponsor: a scoping conversation before we start, the readout at the end, roughly two hours in between.
  • The function head: two working sessions, at the start and the end of week one.
  • Access to the people who actually run the workflows, for the interviews in week two.
  • No data project first. Data readiness is part of the work, not a condition for starting it.

And then what

The diagnosis is where you start. Staying through adoption is why it is worth starting with us.

A diagnostic vendor hands you the answer and leaves. The value in a workflow redesign is not created when the analysis is finished — it is created when the redesigned work becomes the way the organization actually operates, which is months later and several handovers away.

So the same people who read the work stay through building it, communicating it, and embedding it. That is one accountable party across all four stages, and it is the reason to buy the diagnosis here rather than somewhere cheaper.