AI that shows up in EBITDA.

Assessing a target's AI in due diligence means checking two things: whether the company's own AI claims hold, and whether AI threatens its business model from the outside. The first is tested in the data, in who actually built what, and in what runs in production rather than in a demo. The second is tested in what the company sells, and how much of it a customer or an AI-native competitor could now do more cheaply. Before signing, Ampliro scopes the assessment to the time the deal allows. The broader work on processes belongs in the first hundred days.

An AI due diligence checklist for a target should cover six areas: the AI claims in the management presentation and the evidence behind them, the quality and ownership of the data, the AI tools and models in use and the terms they run on, the exposure of the business model to AI-driven price pressure or new entrants, regulatory exposure under the EU AI Act, and whether the organisation can actually adopt AI, from leadership to the people doing the work. The checklist matters less than the judgement applied to it: the same finding can be a red flag in one business and irrelevant in another.

Private equity firms mainly use AI in portfolio companies to lower cost, by taking manual work out of processes such as finance, customer service and reporting, and to lift revenue, through faster quotes, sharper pricing and more time spent selling. Few see it in the numbers. Three in four of the companies McKinsey counts as AI high performers have fundamentally redesigned their workflows, against one in four among the rest. The difference lies in redesigning the work, not in buying the tools.

An AI value creation plan contains the processes where AI should make a measurable difference to cost or revenue, the baseline each one is measured against, the changes to workflows and roles, the tools for which people, and the person who owns each line. It also says what not to do. A plan that adds AI on top of unchanged work rarely moves the numbers.

The ROI of AI in a portfolio company is measured like any other value creation lever: a baseline before the change, agreed measures per process, and reporting over time. Ampliro sets the baseline before anything changes, so the effect of AI can be separated from everything else happening in the company, such as price changes or new hires. The measures are the ones the business already runs on, for example hours per case, lead time, conversion and margin, and they are reported to the board in the format the value creation plan already uses.

AI change management matters more than the tools because the tools are the same for everyone. What separates the companies that see results is whether the work is redesigned around them: which steps disappear, which roles change, and how people learn the new way of working. Companies that add AI on top of unchanged processes pay for licences and see little in the results.

The EU AI Act is a due diligence item, but usually a small one. Most of what a portfolio company runs carries modest obligations. The exposure worth checking is specific: AI used in recruitment or for creditworthiness is high-risk under Annex III, with requirements from 2 December 2027, and AI built into regulated products falls under Annex I from 2 August 2028.

Yes, Ampliro works with investors and portfolio companies in the Nordics and internationally, in English, on site where it matters and remotely where it does not.