AI for private equity, fund and portfolio.

The highest-value diligence use is speed on unstructured material: contracts, customer correspondence, support tickets and technical documentation reviewed in days rather than weeks, with citations back to source so findings are verifiable.

The second is AI-readiness scoring of the target itself. Data accessibility, integration surface and engineering capability determine whether the AI value-creation thesis in the investment committee memo is realistic or aspirational. Scoring this before close prevents 100-day plans that cannot be executed.

The third is competitive exposure. Understanding where a target's category is being disrupted by AI, and whether the target's moat depends on work that is becoming cheap, belongs in the thesis rather than the first board meeting.

Cost-to-serve is usually the fastest lever. Support deflection and agent copilots produce measurable handle-time and containment improvements within a quarter, and the baseline is already instrumented in most contact centres.

Commercial productivity follows: sales copilots grounded in the portco's own pricing, product and account history, rather than generic assistants that the team abandons after a month.

Back-office automation — FP&A close acceleration, procurement and contract intelligence, invoice and claims processing — tends to be less visible but highly repeatable across portcos, which is what makes it attractive at portfolio level.

Demand forecasting and inventory optimisation move the most money in asset-heavy portcos, but require the cleanest data and the longest runway. They belong in year one of the hold, not the first 100 days.

Both, in sequence. Use cases must be shipped inside individual portcos, because that is where the data, the workflow and the P&L sit. But the plumbing underneath — model gateway, retrieval, evaluation, observability, governance — should be built once at fund level and consumed by each portco.

That split gives the fund repeatability without imposing a single roadmap on companies with different operating models. The playbook, not the product, is what transfers.

It also improves exit narrative. A portco running production AI on a governed platform, with measured business impact, is a materially different story at exit than one running pilots.