Energy services group gets an AI roadmap built to withstand investor scrutiny

A private-equity-owned energy services company asked a plain question: where can AI create value for us, and what will it take to get there? Antire’s AI readiness assessment gave the leadership team and the owner a prioritized roadmap of 26 AI use cases and eight foundational programs, and a clear account of the foundations that had to come first.

When the ambition is clear but the foundations are not

The company had just changed hands, and the appetite for AI was already there. Its new private equity owner saw data and AI as central to the value creation plan, the CEO used generative AI daily, and regional leaders had started exploring tools on their own initiative. The next step was to turn that energy into a shared plan. AI strategy, ownership and budget still sat at group level, and the operating companies ran on different ERP systems, with a group-wide replacement planned over the next three to four years.

The owner wanted a prioritized set of AI initiatives it could fund. It also wanted the diagnosis behind them: which use cases the data could support, and which would stall. Antire was asked to run an AI readiness assessment that answered both questions, and to do it in a way that would survive scrutiny from an investment committee.

Evidence before opinions

The assessment was built around Antire’s six-pillar AI readiness framework and a five-level maturity scale from ad hoc to optimized. The evidence came from more than 20 structured interviews covering the CEO, vice presidents, regional general managers and operations managers, and the finance, commercial, HR and IT leads, across five regions and group functions. A company-wide survey ran in two languages. It avoided the term AI and asked about “smart computer tools” instead, so that worries about job security would not color the answers. Three document analyses added the company’s own record: the lessons-learned report from an earlier ERP program, the group AI policy, and a bundle of monthly operational and board reporting.

Every maturity score had to trace back to named evidence. Findings confirmed by three or more independent voices in different roles were stated as fact. Two-witness findings were stated with the witnesses identified. Anything heard once was treated as a hypothesis to test. The owner had set the bar explicitly: a generic “we don’t trust the data” was not a reportable finding unless it said what specifically was not trusted, why, and whether that was still true today.

The answer: fix the base, then build

Maturity varied across the group in ways a single score would have hidden. The most established region had run the same ERP for many years, closed its books quickly and already used analytics in daily decisions. Other regions were earlier in that journey and still relied on lighter, largely manual tools for planning and record-keeping. Confidence in data also differed by type: finance data was well trusted, while operational data was less so.

Some themes applied across the whole group. Operating cost per asset, the figure the owner most wanted for fleet decisions, was not yet captured consistently. Much of the operational know-how sat with experienced, long-serving staff, which made capturing it a priority. Management teams spent two or more days a month assembling reports, time the roadmap aims to give back. An AI policy was already in place, and the next step was to support it with clear ownership, an approval process and training for the many leaders who were already using generative AI.

The most encouraging finding was that in most regions the data already existed and was largely correct. It sat across systems that did not yet share a common data language, so the recommendation focused on making the data easy to access and query. That is a more targeted and achievable program than a broad data clean-up.

A roadmap sequenced around what the data can support

Antire translated the diagnosis into a register of 26 AI use cases and eight foundational programs, each scored for value, cost, feasibility and dependency, and sequenced across Now, Next and Later horizons. The Now horizon held quick wins that run on existing data and tools, such as document and certificate retrieval, tender drafting, report assembly, and the first steps toward a single view of fleet and customer data. The Next and Later horizons held the heavier use cases, including demand forecasting and a useful-age figure for every asset that the owner can trust, each gated to the data accrual and ERP milestones it depends on. Where a use case could not be brought forward with more money or more people, the roadmap said so.

A companion process efficiency register cataloged 80 manual or repeated processes surfaced during the assessment, from monthly consolidation to document retrieval, and showed for each one whether the roadmap covered it fully, partly or not at all. It gave the leadership team a concrete map of where clerical effort sits, and it showed where the roadmap did not yet reach.

An interactive dashboard let the owner explore the register, the maturity scores and the sequencing without wading through a workbook. The investment case followed the same discipline as the findings: a modest first phase of quick wins and two foundational programs, then a larger data platform program staged across six gates, each scoped and priced before any commitment.

Built for an owner who reads every number

This owner reads every cell and had asked for the rationale behind every estimate. Antire reconciled every value figure across the register, the dashboard, the investment case and the executive deck, so the same number appeared everywhere. Indicative value ranges were kept visibly separate from anything delivered. Exit-valuation effects, which are one-off, and annual savings, which recur, were reported separately and in different units. Where the evidence was thin, the register said “sizing in progress.”

The same care went into protecting the people who had spoken candidly. Interview findings were de-identified systematically, down to removing role-and-region pairings that could point back to a single person, before anything reached the owner.

“We could have handed over a long list of exciting AI use cases,” says Øyvind Dale Spørck, Chief AI Officer at Antire. “What the owner needed was to know which of them the data can carry today, which need foundations first, what each step costs, and why we believe the numbers.”

The executive readout to the leadership team and the owner opened with the diagnosis and the maturity picture, moved through root causes and implications, and only then presented the prioritized initiatives with cost, value and the basis for each estimate. It gave the owner and the leadership team a common starting point for deciding what to fund first.

Co-owned by the people who have to live with it

The findings were worked through with the two internal sponsors, the operations lead and the head of digital development, until they co-owned the conclusions. The sponsors pushed for blunter language where the first drafts were too polite about the gap between the most and least mature regions. They wanted the board to see the real state.

“We needed someone to say clearly, in language the board would read, what the real state of our data and processes is,” says the company’s head of digital development. “The assessment gave us that, and it gave us a sequence we can defend.”

What’s next: from roadmap to everyday capability

The first move since the assessment has been enablement. The company has rolled out supported generative AI tools in tiers, with Microsoft Copilot as the company-wide default and Claude for the leadership and management tier. Leaders who had been exploring tools on their own now have supported ones. Antire has been asked to deliver a global AI enablement program around them: a common AI foundation course on responsible use and prompting, tool-specific courses for each tier, departmental coaching on real work, and a network of internal superusers and champions who support colleagues after the program ends.

The foundational programs, starting with a group data platform and data governance, are the next decisions. Each one is scoped and approved on its own evidence, so that the use cases with the largest value, including the fleet analytics the owner cares most about, arrive when the data can support them. The group now has a shared, evidence-based view of where it stands and a sequence it can fund with confidence, so each investment lands on foundations ready to carry it.