AI rollups are real - but so is the hype surrounding them! Insights from our debut “AI in Rollups” Deal Summit

Disclaimer: Unless noted otherwise, views and analysis expressed here are the author's own and based on public sources. The article is intended for informational and entertainment purposes only. This is not financial advice. Please consult a professional for investment decisions.

*********************

If anyone needs proof that AI rollups are all the rage, consider the fact that we had a 90% turnout on the day when a Tube strike paralysed London transportation. Over an afternoon in East London, we went over topics such as:

1. How VCs decide on which industries are suitable for AI rollups, and which rollups are worthy of their time + money

2. Why a bit of HoldCo heresy fuelled Hawk Infinity’s growth to $160M EBITDA 

3. How Unaric is adapting clunky Salesforce applications to the Agentic AI era

4. Why Dwelly and Ralph are obsessed about proprietary tech stacks - and relaxed about departing business owners

Kudos to our longtime supporters Grata / Sourcescrub and PPH Financial. It was equally great to see new and existing vendors join us for the event. Jason Ing from Amplius Partners delivered an excellent “lunch and learn” negotiation seminar. As ever, Caroline Urban from MBM Commercial was generous in sharing insights gained from advising many a serial acquirer. Finally, Stan Beremski from SoundSenseAI is fast emerging as the go-to AI implementation partners for SME acquirers.  

From this month, we are excited to welcome a new strategic partner: Reef Pass Investors.

Reef Pass is a US based investment partnership that specialises in launching serial acquisition platforms, HoldCos, and Long-Term Hold (LTH) platforms focused on making acquisitions across lower middle markets. If you are raising, DM us and we will happily screen your idea on behalf of Reef Pass! 

1) Panel: “Beyond hype: how top rollup investors select AI-first platforms”

To kick things off, our excellent moderator Suzy Smith (Managing Partner at the consultancy RPL) hosted Cyrus Hessabi and Sahil Patwa, the VCs from OpenOcean and Unbound, respectively. Just two weeks ago, I sat down with Cyrus and Sahil to work out what AI rollups actually mean. They had to do it again, but this time live, and in front of a demanding crowd of almost 200!  

Some snippets:

Fundamentally, AI rollups are technology companies capable of carrying out many of the tasks of a typical services business. In industries where it's difficult to acquire customers organically (due to e.g. low propensity to adopt new technology / switch providers), acquiring incumbents makes more economic sense. But only if you are prepared to reinvent the entire business model - starting with the underlying data stack that helps orchestrate the operation - rather than updating the model, one function at a time. When evaluating emerging AI rollups, investors should be asking themselves a question: are these founders merely “sprinkling” automations on top of an existing job? What is the proof point? You must be able to automate core processes.  

The panelists were at pains to emphasise that, contrary to their PE “cousins”, the investment case for AI rollups is much less driven by multiple expansion. Surely this logic would preclude Cyrus, Sahil and other VCs from investing in “off the shelf” rollups that combine third party technology with acquired distribution. If the groundbreaking tech is widely available, what’s the moat? 

Sahil then walked us through his framework for identifying industries suitable for AI rollups:

• Firstly, pick industries where distribution patterns cannot be easily disrupted. To use residential lettings as an example, landlords typically change tenants every 12-24 months. For a lettings agent to land a new customer, the window of opportunity is very short, at most a few weeks long every couple of years. A secondary industry filter, suggested by Suzy and Cyrus, is the measurable impact of AI adoption.  

• The second factor is capital structure related. Since you will be debt funding acquisitions, pick businesses with stable revenue profiles. Avoid cyclical industries, like recruitment (note from Alex: agree, but doesn't mean recruitment rollups are a non-starter - read our deep dive on Kernel Global). Having posted phenomenal revenues in 2021, recruiters were hard hit in 2022 and 2023 once redundancies picked up. Similarly mortgage broking is highly dependent on macro and interest rates

• The third factor is the degree of market concentration. An industry where 10 players represent 80% of the TAM isn't particularly suitable for AI rollups, since you'd be compelled to splurge on a large acquisition from the get-go. Conversely, extremely fragmented industries are challenging because the targets are so small. Sweet spot: industries that have between 1,000 and 5,000 players. 

 Now, this would not be a consultant moderated panel if we didn't touch on the risks inherent in AI rollups. Apparently there are 2 main ones:

• People and change management. Less in the sense of IT system replatforming, and more about convincing people who have not been exposed to powerful technology to embrace new ways of working. Imagine telling a 30 year old company veteran that 80% of his/her job will be done by tech. As Sahil put it, “You will have people wanting to leave, and to take business with them”. One solution is “driving the culture of constant experimentation” (note from Alex: might be a good time to retrain as an executive coach!) 

• Regulation and compliance risks inherent in data aggregation. One mitigant is altogether staying away from over-regulated industries such as healthcare or contract research / clinical research. In Sahil’s opinion, “some of the best founders spend 30-40% of their time talking about risks. What scoring models do they use to get comfort that the customer experience is not going to get affected”. 

My favourite question came at the very end: for the “build, then buy” rollup category, how do VCs validate that the tech does what it says on the tin? According to Sahil, “nothing beats having something in production” i.e. a working prototype already serving end customers.  

2) AI rollup pitch: Unaric

Up next was Peter Lindholm: a serial entrepreneur who co-founded Unaric in 2023 as a Salesforce “platform play”. Fast forward to 2025, Unaric has acquired 20 applications, giving it a customer base of 2,500. 

Why is Unaric worth keeping an eye on? 

First of all, the Salesforce app ecosystem is HUGE. A $35B TAM - one of the largest SaaS ecosystems globally. While Salesforce is used by nearly 90% of Fortune 500 companies, the ecosystem is both hyper fragmented and “very far from being AI ready”. Secondly, the multiplier effect is tremendous. Every $1 in Salesforce revenue generates $6 for the ecosystem! 

Unaric is creating a single Salesforce suite with all the applications that customers need. This is a major value prop: instead of integrations that take 3-4 months, Unaric products can be rolled in a matter of hours. The customers benefit from a single contract and a single point of contact with the vendor. 

In 2.5 years Unaric has built a solid deal funnel: 600 targets reviewed and 20 acquired.

Included within its portfolio is a Payments business that handles $2B GTV. As Peter described, pre-acquisition Payments used suffered from patchy customer service and failed payments. By deploying AI to these issues, Unaric accelerated growth from 0% to 50-60% YoY. One use case was predicting failed payments ahead of time based on customer behaviour, the credit card used, the time of the month etc. 

What does the future look like? 

According to Peter, fewer human and more agentic users. Unaric’s responses, mirroring Salesforce’s strategy, is shifting from seat-based pricing to usage-based pricing. Investing in API connectors. Also: doing more of the same. Buying small software businesses, optimising them, installing AI where appropriate - and harvesting the resulting growth and cash flow (Unaric sports 89% gross margin). 

3) Fireside chat 🔥 Alex Prokofjev x Joakim Karlsen, CEO at Hawk Infinity Software

Joakim kicked off the chat by recapping the amazing story of Hawk Infinity. What started out as a multi-family office in 2016 with a $10M endowment ($5M from Marianne and Johan Michelsen, plus $5M from a property developer seeking diversification) first grew to $50M over 4 years (2016-20), and later to $1B+ as Hawk embarked on a compounding strategy. 

For an in-depth review of Hawk’s business model (including what companies it has bought and for how much), check out our primer on the company and the interview with Joakim.   

Today, Hawk’s portfolio comprises 75 businesses - most of them software - generating $160M in pro forma EBITDA. Overseeing this mini empire is a central team of 20. Joakim is confident that Hawk can add between $50M and $100M in EBITDA per year, mostly from M&A. 

For a casual observer, Hawk may seem like the archetypal Scandi HoldCo. It is not, for it won't sing from the same hymn sheet of “stable share count”. Hawk is famous (notorious?) for requiring sellers to roll over big chunks of sale proceeds into TopCo equity.

Joakim’s counter?

Subscribe to Premium to read the rest.