A national power-generation services operator decides where AI helps its experts

Plant health reviews depend on a few senior experts

The operator’s senior experts review the health of the plants and assets it serves. Much of their time goes to requesting documents and comparing them against the standard, before they reach the judgment only they can make.

When a few people carry that work, their hours set how many reviews the business can take on and what each one costs.

Its leaders chose to map the work before funding a build

The operator’s leaders decided to map the work first, one business line at a time, starting with asset reviews. They used Helios 2, a mapping engagement that starts from business goals and the real work and picks the tool last.

The people who do the work described it: the documents they gather, the steps they follow, the systems they use, and where the time goes.

Each task was scored on its value and on what it takes to build. Then it got one of three answers: a process that stays with people, a workflow with fixed steps, or an agent whose drafts a named person reviews and owns.

Every task on the map has an answer and an owner

The first two business lines produced 54 opportunities across 12 functional areas, and each area has a named owner on the operator’s side. Of the 54, 16 are workflows that need no model, and 5 are processes that stay with the operator’s people.

The operator’s executives and subject-matter experts confirmed the inventory and agreed how each business line and function will be measured. Low-value ideas were cut before scoring, so the map holds what survived.

The experts make the call with the evidence already assembled

The operator’s review work follows one design rule: every score and finding links back to its source evidence and to an expert’s approval. AI drafts, and the judgment on plant health stays with the expert.

The aim is for the experts to spend fewer hours gathering evidence and more on the judgment the plants depend on.

This story covers the map. Anything built from it gets its own scope and price, and is measured against the KPIs the operator agreed.

Its leaders hold a plan they can fund one wave at a time

The first wave holds 33 opportunities that are ready to start now. Another 18 follow in the next wave, and 3 come later. The waves follow each row’s value, build effort, and dependencies, so the first spend goes to valuable work that is ready to build.

Each row carries a named owner and a measure the operator’s executives agreed. The first wave can be funded and tracked on its own before the next one is committed.

The workbook belongs to the operator whether or not it builds with Ludia. The map grows one business line at a time.

Ludia’s part: senior consultants ran the interviews and the scoring with the operator’s teams in the room. They used Helios 2, the method Ludia ran on its own firm first, and the operator’s leaders made every call on the map.

For your IT team: what each row of the map carries

Every row carries a draft operating brief: objective, trigger, inputs, outputs, and guardrails. The map stays tool-neutral until the last step, and each row gets the simplest layer that meets the need. Workflow rows usually land in Power Automate↗; the rest land in Dynamics 365’s own agents, Copilot Studio↗, or Microsoft Foundry↗, with Agent 365↗ governing them once they are live.