AI that pays backinside the hold.

Selling for more than you paid isn't enough any more: returns have to come from growing profit while you own the business. We build AI that takes cost out of your portcos and speeds up add-ons, in weeks, owned by the portco and ready for diligence.

Duncan Anderson

Duncan Anderson, our co-founder, led our work with Vespa Capital, a PE firm: finding where AI would pay back across its portfolio, testing the best ideas in short experiments, and taking four into production. Duncan on LinkedIn ↗

What we see

Where the plangets stuck.

The plan is usually right. Doing it is the hard part.

  • Profit has to grow while you own it

    Bain says a deal now needs 10–12% annual EBITDA growth for the return 5% used to deliver. With holds running about six years, that growth has to come from how the business runs.

  • Add-ons are slow to pay back

    Buying and merging smaller businesses is now how most PE firms grow a portco, but only 25% see results within 12 months. Each add-on arrives with its own systems, customers and pricing.

  • Diligence finds the gaps you didn't

    At exit, the buyer checks whether the business follows its own policies and meets its obligations. Gaps found late come off the price or end up in escrow.

Where it helps

What we can buildfor your portcos.

  • Find where AI pays back in each portco

    We run workshops with each portco's leadership to find the work worth automating, then test the best ideas in short experiments before anyone commits to a build. We did this across Vespa Capital's portfolio: at Acacium, AI now matches clinicians to late hospital placements.

    Worth: Money goes on builds that have already shown they work.

  • Answer the routine customer enquiries

    AI answers customers from the business's own data, such as live prices and bookings, and passes anything tricky to a person. iQ Student Accommodation, a Blackstone company, handles about 100,000 fewer support emails a year this way.

    Worth: More customers without hiring at the same rate.

  • Draft the documents that take weeks

    AI agents draft and check long documents against the business's own standards. Cognite, a Vespa Capital portco, uses AI to write its proposals and check medical claims.

    Worth: Senior people check the work instead of writing it.

  • Use AI on sensitive data, safely

    A layer in the portco's own cloud strips names and personal details out before anything reaches an outside AI service, and puts them back in the answer. Acacium, a Vespa Capital portco, uses AI across its operations this way.

    Worth: AI in regulated portcos without a data protection problem.

    Read the Acacium case study →
  • Get ready for diligence early

    Whereas, a tool we built, maps the business's policies against the regulations and obligations it has to meet, so gaps show up while you own it, not when a buyer looks.

    Worth: Fewer surprises once diligence starts.

Proof

Already working inPE-backed businesses.

  • Vespa Capital: from workshops to AI in production

    Vespa Capital is a PE firm. We assessed where AI would pay back across its portfolio, tested the best ideas in short experiments, and took four into production at two healthcare portcos, with their own teams trained to run them. At Acacium, AI matches clinicians to late hospital placements, and an anonymisation layer lets it use outside AI services on healthcare data. Cognite uses AI to write proposals and check medical claims.

    Read the Vespa Capital case study →
  • iQ Student Accommodation: Q, the student support assistant

    iQ is a Blackstone company. Q answers students using live prices, availability and maps across about 80 buildings and passes anything tricky to a person. iQ now handles about 100,000 fewer support emails a year.

    Read the iQ case study →

The question investment committees ask

What happens to the AI when you sell?

A buyer will ask who owns the system, where the data sits and what it depends on. An AI tool rented from a vendor on a contract the portco can't control becomes a question in diligence rather than an asset.

We build systems to transfer with the business. That keeps them on the right side of the data room.

  • Runs in your cloud

    The system and its data sit in the portco's own infrastructure, not a vendor's.

  • You own the code

    Nothing to renegotiate at change of control.

  • Switch models without rebuilding

    If an AI provider changes its terms or prices, the system moves to another model.

How sovereign AI works →

What happens next

From one portcoto the whole portfolio.

  1. A one-day workshop

    We work through one portfolio company's value creation plan with the operating team and pick where AI would pay back first.

  2. A working prototype in two weeks

    Built on the portco's own data, so the operating partner can judge it on real numbers.

  3. Build it in our AI factory

    We take the prototype to a production system the portco owns.

  4. Into production, then the next portco

    What we learned goes into the next build, so each one is faster.

Building value in a PE-backed business?Let's talk.

Tell us about one portfolio company and the part of the plan that's slipping.

Talk to Duncan

Duncan Anderson

Co-founder, Barnacle Labs · Former IBM CTO for Lloyds Banking Group