A two-lever playbook for PE operating partners: redesign one department in one portco, put the India team inside the portco’s entity, then carry the method to the next company.
Many of the PE portfolios we see are paying vendor margins on operations they could own. Same back-office cost line in every portco, same vendor invoice structure, same compounding leak. The playbook below is the dual-lever approach to fixing it. AI-native org chart on top. Owned India team inside the portco entity underneath. Together they are designed to make each portco’s back office cheaper to run and easier to verify. Whether a buyer also pays a different multiple for that is the buyer’s decision, not a deliverable.
The pattern across the portfolio
Walk through the portcos in a healthcare services or specialty distribution fund. Look at the income statement. In the funds we have looked at, the back-office line is a similar share of revenue in each portco. The vendor names are different. The structures are identical: third-party processing, per-transaction or per-FTE pricing, multi-year contract, modest annual margin compression that the portco celebrates as cost savings.
Now look at where the AI conversation is happening in each portco. It is happening with the same vendor. The vendor is offering an "AI roadmap." The roadmap runs on the vendor’s infrastructure, trained on the combined data of all of their clients. Depending on the contract, the portco may be funding the education of a system it does not own. Three years in, the operational intelligence the portco’s team produces every day (claims patterns, exception logic, customer-specific workflows) can accrue to the vendor’s platform. Read the data and IP clauses before assuming either way.
This is the leak. It is the same leak in every portco. And it is invisible on the income statement until exit, when the buyer’s diligence team asks who owns the AI layer and the answer is "the vendor." That answer weakens the operating story. Sometimes it kills the deal. The playbook is about closing the leak, portco by portco, on a timeline that lines up with the hold period.
Lever 1: the AI-native org chart redesign
Lever 1 is the work of redesigning each portco’s operating org chart in a fixed order: rules-based automation first, AI for suitable unstructured work, people for judgment and accountability. Not retrofitting AI into an existing structure. Redesigning the structure so that Virtual Employees, AI workflow roles with named human owners, handle the routine work and a smaller, more senior human team handles the calls that require judgment, context and accountability.
In an illustrative composite, a 30-person back-office function is redesigned around 6 senior human roles plus 14 governed AI roles: an 80 percent reduction in human roles, with comparable throughput as the design target rather than a guaranteed result. The humans are not survivors of a layoff. They are a deliberate selection: one director, one manager, four senior individual contributors. The Virtual Employees handle first-pass classification, reconciliation, exception triage, and the routine documentation work. Each Virtual Employee has persistent memory across the workflow, defined governance with human approval at the decision points, and audit logs that survive the engagement.
This is the lever that changes the operating story. The banker’s description can move from "this portco saves money on labor" to "this portco runs a documented operating model with governed AI roles and named owners." Some buyers value that differently. We do not quote a multiple for it, because that depends on the sector, the period and the comparable transactions, and none of those are ours to deliver.
Lever 2: the COPO entity inside each portco
Lever 1 alone is not enough. If the AI runs inside a vendor-managed operation, the residue can still leak. The Virtual Employees produce intelligence that may land in the vendor’s data pipeline, and the operating story does not survive diligence.
Lever 2 is the owned India team inside the portco’s own entity. COPO means Company-Owned, Partner-Operated. The portco owns the legal entity, the employees, the data, and the governance. Reliable Group operates the day-to-day: hiring, training, facilities, payroll, compliance, retention, the India labor law paperwork, the technical infrastructure that runs the Virtual Employees. The portco’s client-specific workflows, data and documentation are its own. Trace is Reliable Group software, licensed under the engagement; third-party tools keep their own license terms.
The team inside the COPO entity is smaller than what the legacy vendor structure required, because the Virtual Employees are doing the routine work. The team is also more senior. The senior India hires are exception-handlers and judgment-layer operators who know the portco’s domain better than anyone outside the original US team did. Eighteen months in, those operators are an institutional asset.
Sequencing across the portfolio
Do not start with the largest portco. Pick the pilot against five criteria. Stable management that will still be there in a year. A sponsor with authority to make the decisions the work surfaces. One department with a clear first queue, not a whole function. Clean enough data and system access that a discovery can actually measure. No transaction inside the next year, so the work has time to produce evidence. The first portco is the proof point that funds the conversation in the next.
The Prove-Expand-Compound sequence runs the first portco through three phases. Prove: one workflow scoped tight, with the India team starting under the Flexi model in Reliable Group’s entity. Expand: the COPO entity stood up, Virtual Employees scaled in alongside the team. Compound: year two and beyond, when the institutional residue starts to show up as a quality ratchet the buyer’s banker can defend.
A stage gate, worked through
This is an illustration of how the gates run, not a report of a specific engagement.
Gate one, after the discovery: decide whether to sign the fixed-price proposal for the first build. Evidence: the workflow map, the ranked candidates and the price. Gate two, after the first build has run in production for an agreed period: decide whether to expand to a second workflow or stand up the India team. Evidence: measured hours against the discovery baseline, exception rates and run cost against budget. Gate three, before a second portco starts: decide whether the method transfers. Evidence: what was reused from the first portco and what had to be rebuilt.
Decision rights sit with the client at every gate. We recommend; the portco CEO signs for the portco, and the operating partner decides whether the program moves to the next company. Each later portco gets its own discovery and its own scope. What carries over is the method, the standards and the scorecard, not the design. A sensible pace is one new portco at a time, once the previous gate is passed.
Exit economics
At exit, operating improvement and valuation are different things, and the playbook keeps them apart. Operating improvement is measured: cost to serve, capacity, error rates, documented controls. That is what the work delivers.
A buyer may also read a portco with both levers in place (AI-native org chart, Virtual Employees in production, owned COPO entity) as tech-enabled services rather than a services business. The institutional residue is on the portco’s books, not a vendor’s, and the buyer can take operational ownership of it on day one. But that reading is the buyer’s, and it is not a deliverable. The case to underwrite is the one with no rerating: today’s multiple applied to the improved earnings.
How a buyer capitalizes the story is sector-specific and period-specific, and it is not something the operating work can promise. An earlier version of this article quoted a revenue-multiple lift range here. That figure was not sourced or modeled transparently, so it has been removed rather than re-sourced. What we will say is narrower and defensible: where both levers are real at diligence, the operating claims are verifiable rather than asserted, and a verifiable claim is the precondition for any valuation argument. Whether the market awards a different multiple for it is the market’s decision.
Where the playbook starts: one department in one portco
The playbook starts with a fixed-fee discovery of one department in the first portco: six to eight weeks, ending in a workflow map and a fixed-price proposal for the first build. It is deliberately narrower than the portfolio program. The discovery answers one question for one team; the program, if it follows, is a separate scope agreed portco by portco.
The workflow map is the artifact the operating partner takes to the portco CEO. The method behind it is what the operating partner takes back to the firm: a documented approach the next portco’s own discovery can reuse without rebuilding it from scratch.
Updated November 1, 2026. Earlier versions of this article described a three-to-five-week Blueprint as the entry product and compared exit multiples of 7x and 10x. The Blueprint is no longer offered, and the multiple comparison was unsourced and has been removed.
Start with one department in one portco. The method carries to the next; the scope does not.