Buy, Build, Deploy: Using M&A to Drive AI Transformation — Prathik Naidu, Long Lake

Read the talk

Buy, Build, Deploy: Using M&A to Drive AI Transformation

Title slide introduces the acquisition and AI transformation thesis.Open full source frame
Title slide introduces the acquisition and AI transformation thesis.

Prathik Naidu explains Long Lake’s case for acquiring established services businesses, deploying AI with their employees, and turning saved time into business results.

From a talk by Prathik Naidu

At a glance

Ideas worth remembering

  • Enterprise deployment requires integration, adoption and a way to capture value; records alone do not reveal the judgment and exceptions embedded in everyday work.

  • Acquiring an established service business can bring builders closer to experienced employees, live feedback, customer trust and shared business incentives.

  • The proposed economic gain is operating leverage: serving more demand without labor costs rising as directly with revenue, while maintaining expert service.

  • Time savings are an intermediate milestone. Management must put freed capacity to work before deployment can produce the intended P&L and service outcomes.

The walls between a working agent and a business result

An AI agent can work well on a narrow task and still be far from production. Prathik Naidu, a founding partner at Long Lake, starts with the vendor’s journey: security review, internal coordination and integration with old systems come first. Then someone must improve the interface, train users and decide who is responsible for adoption. Finally, the vendor must choose how to charge, address price sensitivity and demonstrate return on investment.

Why does integration demand so much organizational work? A formal chart might show an owner, managers and teams, but decisions can depend on a web of experienced employees who work across sales and operations and whom the owner consults. An outside builder has to discover those relationships to design something the organization can actually use.

Candidate during the comparison of the formal organization chart and informal decision network; pixels not supplied for review.Open full source frame
Candidate during the comparison of the formal organization chart and informal decision network; pixels not supplied for review.

Connecting a CRM, ticketing system or repository of procedures, contracts and old emails supplies useful records. It does not automatically expose the exceptions, tacit judgment and quick questions directed to an expert. Working with the people who do the job helps reveal what good work looks like when the documented process meets an unusual case.

There is also a difference between creating value and capturing it. In Naidu’s hypothetical, a vendor earns $1 million a year while its client gains $5 million a year in EBITDA. At an assumed 10× earnings multiple, that incremental earnings stream implies $50 million in enterprise value for the client. The multiple and impact are illustrative assumptions, while the vendor’s own contract value depends on renewal and its ability to maintain pricing as alternatives improve.

0:130:43
Suggest correction

This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.

0:13 · section reference included

Buying the business brings access, trust and shared incentives

Building a service firm around AI from day one is a viable alternative: the company can improve an entire workflow, sell the work and own its results. But a new firm must also win customers, establish a reputation and hire specialists. A health system planning a surgery building, for example, may prefer an architecture firm it has trusted for 20 years, with knowledge of its standards and the ability to deliver the full project.

Candidate during the explanation of the ownership flywheel; pixels not supplied for review.Open full source frame
Candidate during the explanation of the ownership flywheel; pixels not supplied for review.

Acquisition is Long Lake’s proposed third route. Ownership brings several assets into the same operating effort:

  • Access: Builders can sit with experienced employees, learn the work and design for a broader service operation.
  • Live feedback: Real tasks and edge cases provide signals about whether an agent got an answer right, which can inform improvements to agents and models.
  • Aligned incentives: The owner and operating team share the goal of improving the business, rather than negotiating a separate vendor’s revenue goals.
  • Trust and scale: An established brand, customer relationships, expert team and cash flow give the effort a base from which to invest.

Naidu’s compounding thesis connects those assets: people and agents work in live operations, the work reveals how the business functions, and outcomes supply data for improvement. Ownership creates the opportunity to keep learning from that work; the gains still depend on building tools that employees use and that improve the service.

5:105:22
Suggest correction

This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.

5:10 · section reference included

The cost of growing a service business

Long Lake focuses on services because the firm sees an opportunity in owning outcomes. The difficulty is that delivery often grows with revenue: an HOA community manager can handle only so many associations, and an architect can design only so many hospitals or schools. More clients generally require more people to hire, train and manage.

Naidu offers a simplified comparison. For an additional $100 of revenue, he says a typical software business might retain about $80, while a service business might spend about $80 on delivery and overhead and retain $20. He calls the latter an 80% marginal tax on growth: even with demand available, scaling requires substantial additional work. These are illustrative economics, not measured margins for a particular company.

What would change if AI gave employees more capacity? Naidu’s proposed chart contrasts labor costs rising with revenue against a future in which costs grow less directly as revenue increases. His thesis is that AI can lower the cost to serve and improve quality while expert employees remain essential. That operating leverage is the desired result of deployment, not an automatic consequence of buying a company.

Candidate during the discussion of revenue and labor costs after AI deployment; pixels not supplied for review.Open full source frame
Candidate during the discussion of revenue and labor costs after AI deployment; pixels not supplied for review.

Naidu describes Long Lake as an applied AI holding company that had partnered with 40 businesses across residential services, architecture, HR and other markets. He reports $7.5 billion raised and a team of 55 across technology, finance and operations. He also describes a recently announced $6.3 billion take-private partnership with Amex GBT as an entry into corporate travel.

9:289:58
Suggest correction

This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.

9:28 · section reference included

Choose businesses accountable for consequential outcomes

As model capabilities improve and the price of intelligence falls, Naidu asks what remains expensive. Long Lake looks for mission-critical services where customers trust a firm with an end-to-end outcome, human judgment matters, and regulation may govern accountability.

Candidate during the comparison of individual tasks and service outcomes; pixels not supplied for review.Open full source frame
Candidate during the comparison of individual tasks and service outcomes; pixels not supplied for review.

The distinction becomes concrete in three kinds of work:

  • Residential services: Drafting a homeowner newsletter is a task; handling an urgent leak is an outcome.
  • Tax: Extracting data from documents is a task; taking responsibility for a complicated tax decision requires judgment and accountability.
  • Travel: Booking a flight is a task; helping a customer through a last-minute disruption is part of delivering the service.

A promising AI application alone is insufficient. Long Lake also looks for strong business and market fundamentals and a team enthusiastic about partnering. Naidu says the firm sees about 5,000 deals a year to maintain a broad funnel and a tight filter.

13:2713:56
Suggest correction

This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.

13:27 · section reference included

The J curve: from investment to capacity to P&L

After an acquisition, the transformation starts with investment. The team builds technology, captures workflows and commits employee time. It must work with the business to iterate and validate the product, account for its distinctive workflows and data, and potentially train models. Naidu uses a J curve to describe this initial effort before the hoped-for payoff.

The next milestone is adoption and capacity: employees use the product and can produce the same output in fewer hours. The J curve answers why that is only an intermediate result. To reach the P&L, management must put the freed capacity to work—for example, supporting growth—and pursue outcomes such as better customer and employee experience and higher revenue per employee.

Candidate during the discussion of capacity and P&L impact on the J curve; pixels not supplied for review.Open full source frame
Candidate during the discussion of capacity and P&L impact on the J curve; pixels not supplied for review.

A shared platform and playbook for both deployment and business transformation could make subsequent work faster. Naidu says Long Lake sees faster deployment and time to P&L impact across its portfolio, though he gives no timings or company-level results for that claim.

15:3216:02
Suggest correction

This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.

15:32 · section reference included

The teams needed to buy, build and deploy

The operating model needs different teams working toward the same outcome. M&A builds relationships and completes partnerships with service businesses. Builders travel to work alongside employees and develop agents for the wider operation. Deployment and operations teams drive adoption and work with management to turn time savings into business impact and a better client experience.

That last step returns to the opening problem. Getting an agent into use matters, but the service business must still decide what to do with the capacity it creates. Long Lake’s model places that decision inside the business it owns and operates.

17:2917:59
Suggest correction

This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.

17:29 · section reference included

Resources

Transcript download
Full transcript

Loading transcript…