Listen to the full episode, and read the interview takeaways, below:
___________
Hugo: You and I both know the importance of value creation within a private equity portfolio. It’d be great to hear your take on where that importance comes in.
Al Ramage: AI is everywhere — discussed and hyped — and in private equity the impact is particularly interesting. Recent years have brought weak exits, fewer IPOs and a higher cost of capital. There’s also a lurking fund-to-fund exit problem that’s parked issues in portfolios, which the weaker macro is now exposing. At the same time, AI is accelerating market change. Heavy usage patterns can leak signals that end up steering big-tech feature roadmaps. Net-net, PE edges towards a more VC-like risk profile where some assets will simply fail rather than underwhelm. The job is two-sided: use AI deliberately for value creation and move fast enough to be the disruptor, not the disrupted.
___________
Hugo: In the market so far, what have you seen work and why?
Al Ramage: I tend to look at AI in six domains:
1) Disruptive business models
2) Customer & revenue especially personalisation
3) Operational excellence & supply chain
4) ESG
5) Governance
6) Digital twins/machine vision and asset optimisation
Lesson one: Don’t treat AI as a tech innovation. Look at the vantage point of business. When the board leads, change pulls through; if tech pushes, people push back. In PE, cashflow is always king. Start where time-to-value is quickest (ops efficiencies, asset utilisation, a customer bump via personalisation/marketing) to create a little “creative cash.”
Execute via iteration: what’s the smallest thing you can do, quickest time to value that shows it works or doesn’t? Even with trillion-scale LLM investment you still need a services layer to adopt/validate with governance alongside, and there’s no way ‘auto complete’ can be left to run a business without a human in the loop.
“Adoption has to co-evolve with workforce skills, governance and reliability — it’s not “open a box and… ta-da, it works.”
___________
Hugo: I've seen from a talent perspective myself, a lot of the private equity firms want to implement AI for value creation, just don't know where to start, who to start with. What would you advise?
Al Ramage: Two places. First, in the portfolio companies: easier to direct change and easier to get someone else to pay. Second, in the fundco/holdco: harder, because it forces PE to “look in the mirror,”deal sourcing, due diligence, decision optimisation. The fastest, most assured path to value is to use a majority of market data and market tools like OpenAI, then seed with a little in-house data in a secure lakehouse/warehouse. Far quicker than trying to perfect all corporate data first. Don’t wait to “perfect” every corporate dataset before you begin.
___________
Hugo: Data is the crux of AI. Across a portfolio with different industries and challenges, how do you really personalise at portco level?
Al Ramage: You need less first-party data than you think. You can buy or access market datasets (free, pennies or paid) and if you’re smart about selection, infer a lot without big spend. In lockdowns, combining Google Maps footfall with scraped company presentations/accounts was “utterly transformative” for deciding which retail tenants to support in a property portfolio. Seed with a little of your own data and the signal sharpens.
You can model per-company or aggregate across a theme; that’s a risk/governance choice. When history is thin or sensitive, use synthetic data (e.g., simulate millions of interactions to learn pricing where your history is one price point).
Pattern: see what exists → combine → test fast → scale. And keep privacy/ethics non-negotiable: be clear on lawful sources, consent and leakage risks as you grow.
“If we drew this to scale, the market data would be the size of Jupiter and my data would be the size of a tiny little rock.”
___________
Hugo: There’s a lot of hype around GenAI. What’s the best way to harness it correctly?
Al Ramage: Three routes. First, the quickest value is individual augmentation. Let people use public tools to segment customers and tailor content. Second, the “trough of trouble”: when you chain steps into intelligent automation, today’s GenAI isn’t good enough unsupervised; keep human-in-the-loop and watch for data leakage unless your data maturity is high. Third, transformational insourcing: you can ask a partner to “attach electrodes” to your business, but be cautious. Tech whizzes rarely understand the business deeply, and many firms don’t know themselves well enough to ask the right questions.
My steer: start with personalisation and decision optimisation, and let specific groups experiment sensibly so you don’t lose your best people.
“GenAI can’t be left to run a business unsupervised.”
___________
Hugo: Personalisation keeps coming up as a theme. Are we just talking marketing or is this an operating-model shift?
Al Ramage: The old shared-services/outsourcing playbook: standardise, simplify, automate, share, drive cost down but left monolithic processes you can’t personalise. True personalisation is a business-model change. Atomise the process. Think doom-scrolling: tiny next-best-action decisions keep you engaged and monetised. At scale, you need an automation engine making those decisions and fast feedback loops to learn and adapt.
Done well, the gains are orders of magnitude. Once you’ve got that first bit of value, there’s momentum for the next step. And a plea to PE owners and management: don’t stop at customer marketing; you can personalise for staff, stock and assets too.
___________
Hugo: Before we wrap, what top tips would you leave with senior leaders?
Al Ramage: The opportunity is immense, but there are hefty lumps and pitfalls. Start small and get expert advice. Get clear on what aligns with your strategy. Excite employees and customers. Don’t frame it as job-taking; frame it as doing something amazing in the market. Then break it down and iterate.
Deliver small things fast and learn. Partner; capability outside your perimeter exceeds what you can hire. Be astute about who and how you incentivise them. Make risk a live board topic every meeting: Are we taking enough commercial risk on AI? What risks will we never take and what does that constrain?
“Value at risk — keep it small. Time to value — keep it quick.”
Whether you’re building your first AI playbook or sharpening one that’s already in flight, Al’s message to senior leaders is clear: start from the business, not the tech; move fast on quick, cash-positive wins; use market data first and seed with a little of your own; iterate in short cycles; and keep a human firmly in the loop.
Personalisation isn’t a marketing tactic, it’s an operating-model shift that can lift productivity across teams, stock and assets. Boards set the tone: define the risk appetite and red lines (especially around data), then back teams to learn in the open.
At Montash, we place AI Operating Partners™ into PE-backed businesses to lead market-data-first strategies, decision optimisation and personalisation, and to build the AI operating model, governance and human-in-the-loop guardrails around them.
If you’d like to explore whether an AI Operating Partner is the right lever for your portco or fundco; get in touch for more information, here.
Check out our channel to uncover more insights with leading voices in technology.
To get in touch with Hugo Din and learn more about Al Ramage, head to LinkedIn to connect.