Part of our guide: AI in financial services
If your business is PE-backed, AI strategy looks different: you have Value Creation Plan targets, a hold period (maybe five years, maybe three) and investor scrutiny on returns. You also have portfolio partners who might be trying similar things, and that context changes everything about how you think about AI.
The best PE-backed businesses that are winning with AI aren't doing it faster or more aggressively than everyone else, and they're doing it differently. They're thinking about portfolio-wide leverage and not just single-entity wins, sequencing to align with exit timelines and thinking about how AI demonstrates value to the next buyer.
Dan saw this directly as Group CTO at Pinnacle Pet Group, where 12 separate insurance brands sat within a £50m technology portfolio and each one faced the same AI questions. Executing them independently would have been chaos, and getting them to move coherently, share foundation and still maintain brand autonomy was the actual value creation.
The specific PE-backed constraints:
Value Creation Plans have timelines. You can't just meander into AI, and there's usually a clear plan: "We'll reduce cost by X, grow revenue by Y, improve margins by Z." AI gets slotted into that plan, and the question that matters is which initiatives drive the metrics that matter to the VCP.
Portfolio coordination matters. If you have multiple operating companies, they're probably facing similar problems, and each one going rogue with different AI vendors, different approaches and different timelines is expensive and fragmented. Making them all move in lockstep, on the other hand, removes competitive advantage, so what you need is coordination without conformity.
Exit visibility shapes decisions. If you're being sold in three years, that changes what you build, and you aren't building a 10-year AI capability programme. You're building something that demonstrates maturity and value to the buyer, and that buyer is probably a larger company, a strategic buyer or another PE firm. What they care about is repeatable, governable and auditable AI, not research-stage ML.
Timelines are tight. Value Creation Plans usually have 100-day plans, quarterly reviews and annual milestones, and that's not a lot of time for foundational work, so you need to move fast.
The portfolio-wide approach that actually works:
Identify shared problems. If you have three insurance brands, they probably all need claims automation and they probably all need to automate regulatory reporting. Instead of each one building it separately, you build it once, at the centre, and each of the brands adopts it.
That doesn't remove brand autonomy, and what it removes is pointless duplication.
Sequence for VCP impact. The AI initiatives that directly drive your Value Creation metrics come first, and for an insurance portfolio that's probably cost reduction in claims handling or underwriting. For a financial services portfolio it's probably KYC automation or fraud detection, and for a professional services portfolio it might be document processing or client onboarding. Identify the AI initiatives that move your needle and sequence those.
Build repeatable execution. You are doing AI across multiple operating companies rather than once, and by the third or fourth deployment you should have gotten dramatically faster and cheaper. Build the templates, the playbooks and the governance frameworks that you can reuse, and that is how you get portfolio leverage.
Govern at the centre, execute locally. The operating companies shouldn't be making random AI vendor choices, and they should still have flexibility on how they implement and integrate. That means central governance on the approach, the standards and the validation, and it means local execution on deployment and rollout.
Demonstrate value clearly. This is non-negotiable for PE, and every initiative needs a clear before-and-after: cost reduced by X, time saved by Y and revenue increased by Z. If you can't measure it, then don't do it.
The sequencing that works for PE-backed businesses:
Year one: Foundation and obvious wins. Pick one obvious, high-impact AI initiative that aligns to your VCP and delivers clear value within 6-12 months, whether that's claims automation, KYC triage or document processing. Use that first win to fund larger capability-building, building the governance framework, testing vendor approaches and starting to build reusable templates.
Year two: Portfolio deployment. You've learned from year one, so now you deploy the same approach across your operating companies and you should be faster and cheaper by 40-50%. You're probably running two to three major initiatives in parallel by then, refining the governance based on what you learned.
Year three: Scale and second-order gains. By year three you've deployed core initiatives and you're looking at second-order gains, using the foundation to enable more sophisticated AI. You're probably thinking about what this looks like to your eventual buyer.
Years four and five: (if relevant) You're optimising for exit, and the AI capability that a strategic buyer finds valuable has mature governance, repeatable execution, clear cost or revenue impact and auditable decision-making.
What this looked like at Pinnacle:
Twelve insurance brands each needed to reduce claims processing costs, and building it twelve times independently would have cost millions and taken years. Instead his team built a core claims processing platform at the centre, and each brand configured it for their specific workflows. The whole thing cost a tenth of the alternative and took a quarter of the time that the alternative would have taken.
We had a technology strategy group that set standards on governance, vendor validation and model testing, and each brand had freedom on which initiatives to pursue first. That freedom mattered because each of their businesses was different.
We measured ruthlessly on cost saved, time saved and accuracy improved, and every quarter we reported back to the partnership on progress. That regular measurement meant we could course-correct fast if something wasn't working.
By the time the portfolio exited, it had a mature, repeatable AI capability that was genuinely valuable to the buyer, and it wasn't bleeding-edge research. It was boring, effective and governable technology that drove real business outcomes, and that is exactly what strategic buyers want.
The partnership model that works for PE:
You need a partner who understands the PE context and gets that you're thinking about portfolio coordination and not single-entity wins. Someone in that seat understands exit timelines and what buyers actually care about.
A consulting partner should be able to say: "Here's where you have portfolio leverage. Here's where you should centralise. Here's where you should give operating companies autonomy." They should also understand governance for regulated sectors and they should be able to help you measure and demonstrate value.
A fractional Chief AI Officer should be able to work across your portfolio, setting strategic direction, helping with vendor decisions and coaching operating company leaders. That is someone who has actually been in the operator seat at scale.
Get in touch if you're building AI strategy for a PE-backed business and you want to talk through portfolio coordination and execution.