← All Insights

The New PE Operating Team Is an AI Engineering Firm

aiprivate-equityvalue-creationportfolio-operations

Blackstone and Hellman & Friedman just showed everybody where private equity value creation is going.

The announcement is easy to describe. Anthropic, Blackstone, Hellman & Friedman, Goldman Sachs, and a group of other investors created Ode with Anthropic, a standalone enterprise AI services company. The Wall Street Journal reported it as a roughly 160-person team tied to a $1.5 billion joint venture. Blackstone’s May announcement described a new AI-native services firm with Anthropic engineering and partnership resources embedded directly into the team. H&F’s July launch announcement said Ode is built on the foundation of Fractional AI, led by Chris Taylor as CEO and Eddie Siegel as CTO, and staffed by experienced AI engineers and operators.

Most people will read that as another big AI partnership.

This is the operating team being rebuilt in public.

For years, private equity value creation has meant functional playbooks, procurement leverage, reporting discipline, pricing work, sales process cleanup, and the occasional real operating intervention. That model is not going away. But it is incomplete now, because AI is not a software category that sits beside CRM, ERP, HRIS, and finance tools. AI is becoming a new operating layer across the company.

That changes the job.

If you own a portfolio company today, the question is not whether the company has access to a model. Model access is increasingly available. The question is whether somebody can walk into the business, understand how the work happens, find the workflows where time and margin disappear, connect the models to the systems that matter, and rebuild those workflows into something faster and cleaner.

That is not a prompt library.

That is operating work.

The Big Firms Are Building What The Market Will Need

Blackstone and H&F are doing this because they can. They have the capital, portfolio density, executive access, and brand strength to partner directly with Anthropic, build a dedicated company, and push that capability across a large base of assets. The participation of Goldman Sachs and other major institutional investors makes the point even sharper: this is not a science project buried in a value creation deck.

It is infrastructure.

OpenAI’s Deployment Company points in the same direction from the model-company side of the market: enterprise AI work is moving closer to implementation, not farther away. The pattern is no longer theoretical. The largest PE-linked ecosystems are not waiting for mid-market software vendors to figure this out on their own.

They are putting builders in the engine room.

That phrase matters. AI transformation does not happen in a board packet. It happens in the operating guts of the company: onboarding, underwriting, invoice processing, support, reconciliation, collections, compliance review, ERP write-back, quote-to-cash, pricing, and every workflow where a human being is copying information from one system into another and calling it a process.

That is where the money is.

The old software implementation model was basically: select a vendor, run procurement, map requirements, configure the tool, train users, and hope adoption sticks. That can work for stable categories. It does not work well for AI. The real value is not in the model. It is in how deeply the model gets wired into the way the company works.

The important part is the second half.

Deep familiarity with how each business runs.

Why This Becomes Table Stakes For PE

Every private equity firm large enough to have a real value creation function is going to need dedicated AI capability. Some will build it internally. Some will partner. Some will pretend their existing digital team can absorb it and find out the hard way that AI delivery is not the same as software procurement.

The big firms have a natural advantage. They can hire a central team, justify the fixed cost across hundreds of portfolio companies, and get privileged access to the model companies because they represent massive enterprise demand. If Ode learns across manufacturing, healthcare, financial services, retail, real estate, and infrastructure, every deployment makes the next one better.

That is a powerful model.

But it is not available to most funds.

Most private equity firms do not have Blackstone’s balance sheet. They do not have H&F’s scale. They do not have a portfolio services army. Many do not even have a full value creation team in the way the mega-funds do. They have partners, operating advisors, a few functional specialists, and companies that still run on email, spreadsheets, old ERP instances, and manual knowledge trapped in people’s heads.

Those firms are not going to build Ode. They still need the work done.

That is the opening.

The middle market is going to need outside AI operating partners who can do three things at once: understand the sponsor’s agenda, understand the portfolio company’s workflow, and bring real technical capability without turning the work into a twelve-month consulting program.

Former operators matter here. Technical talent matters here. Neither is enough by itself.

If you only bring AI builders, they may build around the wrong workflow. If you only bring operators, they may know where the value is but lack the engineering depth to automate it properly. If you only bring traditional consultants, you get a deck and a road map, and the business goes back to moving data by hand on Tuesday morning.

The winning model is operator plus technologist, sitting close enough to the work to change it.

The Real Diligence Question Changes

This also changes how sponsors should diligence companies.

The traditional question is: what is the business, what are the numbers, where is the margin opportunity, and what can we do post-close?

Those questions still matter. But AI adds a more uncomfortable question: what is the work actually made of?

Not the org chart. Not the systems diagram. The work.

Who touches the transaction? Who approves the exception? Who fixes the bad data? Who answers the customer? Who reconciles the file?

In a lot of middle-market companies, the operating model is people holding the business together with memory, email threads, and swivel chair operations. That may be fine when labor is cheap and the market rewards the existing margin profile. It is not fine when a competitor starts rebuilding the same workflows with AI agents, cleaner data, and fewer handoffs.

The diligence question becomes: how much of this EBITDA is trapped behind manual work that a better operating system can unlock?

That is not an IT question.

It is an investment question.

If a sponsor can underwrite pricing, procurement, working capital, and add-on M&A, it can also underwrite workflow conversion. But workflow conversion requires a different kind of inspection. You have to get below management interviews and watch where time disappears.

That is where most deal teams I have sat across from are weak. Not because they are not smart. They are extremely smart. But many have never run an operating company. They have not lived through the ugly part where the model says the process should work and three humans are still manually reconciling the same file every Friday afternoon.

That is the engine room.

That is where AI either creates EBITDA or becomes theater.

Cost Cutting Is The Small Version Of The Story

The easy version of AI value creation is cost reduction. Fewer people, faster processes, lower service cost. That is real. A sponsor who ignores it is being naive.

But the bigger story is revenue and operating capacity.

Ode’s public positioning is telling. H&F’s July launch language is not only about efficiency. It is about high-priority AI initiatives and systems built around client subject matter expertise. The WSJ reporting also frames the effort around growing portfolio businesses, not simply removing cost.

That distinction matters because PE firms do not win by cutting everything until the company is hollow. They win by making the business more valuable for the next owner. Sometimes that means freeing capacity so the company can sell more without adding the same back-office cost. Sometimes it means launching a product line because the company can finally package data, expertise, and workflow into software.

For payments and financial services companies, this is especially important. The raw material is already there: transaction data, underwriting signals, supplier behavior, customer interactions, exception history, fraud patterns, sales activity, pricing files, disputes, compliance notes, and operating decisions buried across systems. They are short on the operating layer that turns that data into better decisions and better products.

That is why this is so relevant to Fern Capital.

Engine Room is the Fern Capital version of this idea: a focused AI operating transformation offering for PE sponsors and portfolio companies, led from a former operator’s seat and built with technical partners on a project-by-project basis inside the workflows that drive value.

My thesis around the Engine Room AI transformation offering is simple: most PE-backed financial services companies do not need an AI strategy. They need someone to rebuild the work.

The sponsor needs measurable value: EBITDA, speed, revenue, risk reduction, better reporting, cleaner integrations, fewer manual handoffs. The management team needs help that does not bury them under a giant consulting machine. The employees need systems that remove low-value work instead of asking them to become AI experts on top of their actual jobs.

That is the work.

The Risk For Smaller Funds

The risk is that the largest firms turn AI transformation into a structural advantage.

If Blackstone can bring Ode into a portfolio company and build AI products, automate workflows, and improve customer experience faster than a smaller sponsor can, then the gap between mega-fund operating capability and middle-market operating capability widens. That gap already exists in procurement, recruiting, data, executive networks, and functional expertise. AI makes it more visible because the work compounds.

That is what smaller funds have to respond to. The answer is not to pretend they can copy Blackstone dollar for dollar. They cannot. The answer is to build or buy a focused version of the same capability, pointed at the sectors they own and the workflows that move the P&L.

In payments, that means being inch wide, mile deep. Do not build generic AI transformation theater. Build around merchant onboarding, residual reporting, chargebacks, risk review, underwriting, partner support, ERP and ISV integrations, ticket triage, reconciliation, pricing, and customer success. Build where the money moves. Build where the margin leaks. Build where the data already exists but nobody has turned it into operating leverage.

That is how a smaller fund competes.

Not by having a bigger AI brand.

By being closer to the work.

What Sponsors Should Do Now

There are three practical moves I would make if I were sitting inside a PE firm looking at this.

First, inventory the portfolio by workflow density, not software spend. Look for high transaction volume, manual touches, fragmented data, repeated exceptions, and measurable service or revenue constraints.

Second, force every AI initiative to name the operating metric before the build starts. Hours saved. Days removed from onboarding. Tickets resolved without escalation. Applications underwritten per employee. Revenue lift from faster quote turnaround. If the metric is vague, the work is not ready.

Third, put operators and builders together from day one. The operator knows which process matters. The builder knows what can be automated, what needs a human, and what the system can become as the models improve. Separating those two creates the usual failure pattern: business people writing wish lists and technologists building around incomplete context.

The best AI work is not magic. It is a tight loop between the people who know the business and the people who can rebuild the machine.

That is why Ode matters. It is not just a new services company. It is a signal that the private equity operating model is changing. The big firms are formalizing AI engineering as a value creation capability.

The rest of the market will follow.

The question is not whether every fund will have its own Ode. Most will not.

The question is who they partner with when they realize they need one.

This is one of the major work streams I am building at Fern Capital. Engine Room is designed for exactly this gap: PE sponsors and portfolio companies that need AI operating leverage but are not going to build a Blackstone-scale AI engineering company themselves. Fern Capital leads the work as a focused advisory practice, pairing operating experience with technical builders where the project requires it.

More on Fern Capital: https://www.ferncap.com

Sources