Most operating partners have AI in the value creation plan and zero deployable scope at the portco level. The gap is not ambition. It is specificity. A real AI value creation plan turns the line item into three production-ready scopes — sized, sequenced, and mapped to the next 18 months of the hold. This piece walks through the 30-day path used to get there, and the three signals at day 30 versus day 60 that tell an operating partner whether the plan actually deployed or just regenerated itself for the next LP letter.
The forcing function is Bain’s 2026 math. With borrowing costs in the 8-9% range and leverage compressed from roughly 50% of enterprise value in the prior decade to 30-40% today, sponsors now need 10-12% annual EBITDA growth to clear a 2.5x MOIC over a five-year hold — up from the 5% growth that financial engineering and multiple expansion used to deliver. Bain’s shorthand, "12 is the new 5," is now the planning constraint for every portfolio company. (Bain Global Private Equity Report 2026)
That is the math behind the urgency. AI is the line item operating partners are pointing at to close the gap. The question this article answers is how to translate that line item into three deployments the portco CTO can actually scope, fund, and ship.
The honest baseline: the line item is not the plan
Two pieces of data frame the gap between "AI in the value creation plan" and AI that moves EBITDA. McKinsey’s 2025 State of AI survey found that only 39% of organizations report enterprise-level EBIT impact from AI, and most of those report less than 5% of EBIT attributable to AI use. Roughly 6% of respondents qualify as "AI high performers" with greater than 5% EBIT impact. Out of 25 organizational attributes tested, the single largest predictor of EBIT impact is workflow redesign — and only 21% of organizations using gen AI have actually redesigned a workflow. The other 79% are layering AI on top of how the work already runs, which is precisely the pattern that produces the disappointing distribution above. (McKinsey, The State of AI in 2025)
BCG’s parallel research is harder still: across 1,250 companies surveyed in 2025, 60% report no material value from their AI investments, and only 5% capture substantial value at scale. The 5% group is described as having reshaped business processes around AI rather than purchased tools next to existing ones. (BCG, The Widening AI Value Gap, 2025)
Inside private equity specifically, AlixPartners pegs the failure rate higher: more than 80% of AI programs at portfolio companies still fail, usually because of misalignment on which use cases to pursue, weak adoption, and no measurement framework. (AlixPartners, Practical AI for PE Operating Partners) Bain’s GP Outlook adds the operator’s self-assessment: roughly half of GPs report AI initiatives in their portcos are meeting expectations, and nearly 40% do not expect material financial impact from AI in 2026.
The numbers above are not an argument against AI. They are an argument against unscoped AI. The 5% of companies that capture substantial value share one trait: they pick a small number of workflows, redesign the work around the model, and instrument the outcome. The path below is how operating partners get a portfolio company onto that side of the line in 30 days.
Why "AI in the plan" does not deploy itself
Three failure modes account for most of the distance between an AI line item in the value creation plan and a deployment that the buy-side QofE team can validate at exit.
The plan names categories, not workflows
The value creation plan says "AI in finance," "AI in customer ops," "AI for sales productivity." The portco CTO reads that, agrees with all of it, and has no idea what to fund first. Categories do not deploy. Workflows deploy — specific ones, named at the level of "automate three-way match for the AP queue feeding NetSuite at the platform company," not "modernize finance." Until the line item is workflow-specific, no procurement department, no implementation partner, and no internal engineer can act on it.
The portco CTO is on a different time horizon
Operating partners think in hold-period quarters. Portco CTOs default to the timing of their existing roadmap, which usually has 18 months of committed work that has nothing to do with AI. Without an explicit decision to displace something else from that roadmap, the AI initiative sits in "we agree it’s important" purgatory. Sequencing has to be a portfolio decision, not an addition.
No one owns the EBITDA-impact model
The fastest way to identify a stalled AI initiative is to ask who owns the EBITDA-impact model and what the run-rate baseline is. If the answer is "the portco CTO," the initiative is stalled — CTOs do not get paid to defend EBITDA at QofE. The model has to be co-owned by the operating partner and the portco CFO, with the run-rate baseline locked in writing before deployment, or the savings will not survive the buy-side’s validation lens at exit.
None of these are exotic problems. They are the same three coordination failures that show up in every operating-partner postmortem when the LP letter quietly stops mentioning AI two years after it was the centerpiece. The 30-day path below is built around closing them, not around picking a tool.
The 30-day path: what an AI value creation plan actually looks like
The path has four phases over 30 calendar days at a single portfolio company. Each phase produces a tangible artifact that lives outside any single human’s head — transcripts, process diagrams, scored opportunity matrices, written scopes, and a signed EBITDA-impact model. Day 30 ends with three production-ready initiatives the portco CTO can take to procurement.
Operator interviews across 6-8 functional leads
Structured working sessions with the portco CFO, controller, head of customer ops, head of revenue ops, head of supply chain (where applicable), VP of sales, head of HR, and the CTO. Sessions are recorded and transcribed. No surveys. No frameworks handed out in advance. The output is a baseline inventory of 8-12 candidate processes ranked by leadership pain and perceived margin drag — the operator’s view of where the work hurts, captured before any analysis layer is applied.
Transcripts become end-to-end process diagrams
Each candidate workflow is mapped at the level of system, handoff, decision point, and exception. The diagrams force visibility into the cross-functional handoffs the function-level interviews miss — the place where most AI value actually hides. Operators routinely see how the work flows across their portfolio company for the first time. The deliverable is a set of 8-12 process maps in the operator’s own language, not in a consultant’s framework.
Score, rank, and shortlist the top three bets
Each process map is scored on four axes: estimated EBITDA impact range (sourced to comparable benchmarks), implementation complexity, system dependencies (does the existing ERP or CCaaS platform already ship the AI feature, or is integration required), and hold-period fit. The shortlist is pressure-tested against the value creation thesis the sponsor underwrote at deal close. The output is a ranked top-three with projected impact, payback, and the explicit reason the other 5-9 candidates were not selected.
Production-ready scopes plus an EBITDA-impact model
The top two initiatives get deployment-ready scopes — statements of work that procurement can put on a vendor or that an internal engineering team can pick up the next sprint. The third initiative gets a discovery scope sized for the next quarter. An EBITDA-impact model ties each initiative to a specific cost center, a specific run-rate baseline, and a documented validation method. Day 30 closes with an executive playback to the operating partner and portco leadership and a signed memo locking the run-rate baseline before any tool goes live.
Two things to flag about the structure. First, the 30 days produce three scopes, not one. Funding only the highest-ranked initiative is the most common operator mistake; it concentrates execution risk on a single vendor and a single workflow, and when (not if) one of the three slips, there is no second engine. Second, the deliverables are written, not slideware. A scope that lives in a deck cannot be put on a vendor. A scope that lives in a statement of work can.
How to know it actually deployed: signals at day 30 versus day 60
Most "AI value creation plan" engagements look identical at day 30 and diverge violently by day 60. Three signals separate the deployments that compound into real EBITDA from the deployments that produce a deck and an LP-letter line.
Day 30 sign — written scopes the portco CTO can hand to procurement
If at day 30 the artifact is a "roadmap," "framework," or "strategy deck," the engagement has produced theater. The artifact has to be a statement of work specific enough that a vendor can quote it or an engineering team can sprint-plan it. If it is not, day 60 will look like day 30 with more meetings.
Day 60 sign — one workflow already in production behind a confidence threshold
By day 60, at least one of the three initiatives has a model running in production against a clearly defined confidence threshold — touchless invoice processing above 60% confidence, AI-deflected tickets routed to humans below a calibrated CSAT floor, agent-assist suggestions accepted at a measurable rate. If by day 60 nothing is in production behind a real threshold, the program will not produce EBITDA in the hold.
Day 60 sign — the EBITDA-impact model has a baseline the CFO has signed
The QofE team at exit will not give credit for AI savings tied to a moving baseline. The CFO has to sign the run-rate baseline before the deployment goes live, and the variance has to be tracked monthly against that baseline from the first day in production. If the baseline is "we’ll figure it out as we go," the buy-side will figure it out instead, and the operating partner will lose the multi-turn EBITDA contribution at the precise moment it should have shown up in the CIM.
The pattern operating partners describe most often, in retrospect, is that the difference between a successful AI value creation plan and a stalled one was visible by day 60 — nine months before the stall became obvious in the LP letter.
What "scoped" looks like at the fund level: a 2026 named example
The clearest 2026 example of "AI in the value creation plan" being scoped rather than narrated is the multiyear strategic partnership Thoma Bravo and Google Cloud announced on April 15, 2026. Thoma Bravo — with more than $183 billion in AUM — signed a fund-level deal that gives portfolio companies access to Google Cloud’s Gemini and Gemini Enterprise agentic AI platform, forward-deployed engineering teams, and access to Google Cloud Marketplace co-sell motions. The cybersecurity arm of the portfolio — named Proofpoint, SailPoint, Darktrace, Ping Identity, Sophos, Imprivata, and Exabeam, generating roughly $8 billion in combined revenue — is the first wave. (Thoma Bravo & Google Cloud press release, April 15, 2026)
Two things matter about that announcement for a mid-market operating partner who will never sign a $183B-AUM strategic partnership. First, the structure. The deal is not "AI strategy" at the fund level — it is named platforms, named portcos, and named delivery resources. That is the shape of "scoped" at fund scale, and it is the shape every operating partner should be replicating at their own scale: named workflows, named systems, named delivery resources, named impact ranges. Second, the gap it implies. If a $183B sponsor needed a multiyear partnership to make AI compound across the portfolio, the answer at a $1-3B mid-market sponsor is not a strategy deck either — it is the portco-level equivalent of the same scoping discipline, run on a 30-day cadence per portfolio company.
What an operator gets at day 30 versus what a deck gets you
The contrast below is the practical answer to "what does $200K of operating-partner budget actually buy for an AI value creation plan." It is the difference between a scoped engagement and a strategy engagement at the same headline price.
| Artifact | Strategy engagement | Scoped 30-day engagement |
|---|---|---|
| Workflow inventory | Industry-benchmark categories | 8-12 named workflows from operator interviews |
| Process visibility | Reference architecture diagrams | End-to-end maps of the actual workflows |
| Prioritization | Heat map across pillars | Top 3, scored on 4 axes, with rejection reasons for the rest |
| Procurement-ready output | Recommendations | 2 deployable scopes plus 1 discovery scope |
| EBITDA accountability | Estimated impact ranges | Signed run-rate baseline tied to specific cost centers |
| Day 60 state | Steering committee meetings | One workflow in production behind a confidence threshold |
Sequencing: how the 30-day path lands in the next 18 months of the hold
The 30-day path is the front end of a longer arc. The shape of the next 18 months is well-documented in the AlixPartners and Hackett research and converges on the same sequence regardless of vertical.
Run the 30-day path; deploy initiative #1 in finance back-office
The first deployment is almost always finance back-office — AP automation, three-way match, close-cycle automation, or reconciliation. Hackett’s 2025 Digital World Class Finance benchmark documents that top-performing finance organizations operate at 45% lower cost as a percentage of revenue, deliver 74% faster executive insights, and require up to 42% fewer FTEs than their peers. (Hackett Group, June 2025) The reason finance lands first is unglamorous: the work is high-volume and rule-based, the ERP vendors ship most of the AI features natively, and the savings are easy to baseline.
Deploy initiative #2; instrument month-over-month variance
The second deployment is typically customer ops or pricing, depending on the portco’s data quality. Variance against the run-rate baseline is reviewed monthly, signed by the CFO, and rolled into the operating partner’s portco review. The variance review is the artifact the buy-side will eventually trust at exit; if it does not exist by month nine, no narrative at exit will recover the credit.
Deploy initiative #3 and lock the QofE-defensible documentation
The third deployment lands inside an existing operating cadence rather than as a new program. By month 18, the portfolio company is generating an AI commentary that is grounded in cost centers, run-rate baselines, and signed CFO attestations rather than in adjectives. That is the form the LP letter wants and the form the buy-side will validate.
The LP-letter pressure cycle: why the timing matters now
The reason operating partners feel the "AI on every page and zero deployed" gap most acutely twice a year is timing. Quarterly LP statements are typically distributed within 45 days of quarter-end, and the deeper semi-annual and annual letters land in the windows that cluster around late February and late August. That is when the AI commentary in last year’s letter has to either compound — with month-over-month variance against a real baseline — or quietly drop.
An operating partner who runs the 30-day path in early summer has a deployment in production by late summer, three months of variance data by the late-fall LP touchpoint, and a defensible AI section in the year-end letter that is grounded in numbers. An operating partner who waits to start until the year-end letter pressure is acute does not have the runway to produce variance data in time, which is how the quietly-stop-mentioning-AI pattern starts.
What this means for operating partners
Three takeaways for the next quarterly portco review.
First, treat "AI" in the value creation plan as a diagnostic, not a deliverable. If the line item reads at the level of category — "AI in customer ops" or "AI for finance" — the plan is one decomposition pass away from being executable. Run the 30-day path at the most-stalled portco in the portfolio first, not the most digital-mature one. The most digital-mature portco is usually deploying without help; the stalled portco is the one where the operating partner’s leverage is highest.
Second, sequence the run-rate baseline before the tool. Across the McKinsey, BCG, AlixPartners, and Hackett research, the variable that most predicts EBIT impact is workflow redesign and instrumentation, not tool selection. The portcos that miss the AI value creation goal are not the ones that picked the wrong vendor; they are the ones that picked any vendor before the EBITDA-impact model existed. Lock the baseline first.
Third, ship three scopes, not one. A single scoped initiative has too much execution risk to carry the AI line of the value creation plan on its own. Three scopes — one in finance back-office, one in customer ops or pricing, one in a portco-specific category — give the operating partner a portfolio of bets at a single portfolio company, which is the same risk discipline that runs at the fund level.
The 30-Day Path, Quantified
How Proactive Logic helps with this
Proactive Logic runs the 30-day path described above as a fixed-fee, forward-deployed engagement at a single portfolio company. The deliverables are written, not slideware: 8-12 process maps in the operator’s own language, three production-ready scopes, and an EBITDA-impact model the CFO can sign. Three productized engagements map to where an operating partner is in the cycle:
- The AI Value Creation Sprint — the 30-day path itself, run at one portfolio company. Function mapping, process mapping, AI hotspot analysis, and deployable scopes for the top three EBITDA bets.
- The AI for EBITDA Framework — a structured diagnostic that sizes finance back-office, customer ops, pricing, supply chain, and sales bets against a single portco, with cited impact ranges and 30/60/90-day actions.
- The PE Portfolio AI Readiness Benchmark — a fund-level engagement that scores 5-15 portcos in parallel and produces an investment recommendation for which portco gets the next dollar of value-creation budget.
All three engagements are scoped, fixed-fee, and built around the principle that an AI value creation plan is a decomposition problem, not a strategy problem.
Further reading
- Bain & Company — Global Private Equity Report 2026
- McKinsey QuantumBlack — The State of AI in 2025
- BCG — The Widening AI Value Gap (2025)
- AlixPartners — Practical AI for Private Equity Operating Partners
- The Hackett Group — Digital World Class Finance, 2025
- Thoma Bravo & Google Cloud — Strategic AI Partnership Announcement (April 15, 2026)
- Proactive Logic — AI for EBITDA: The 5 Places It Actually Moves the Number
- Proactive Logic — The Operational Efficiency Imperative
- Proactive Logic — Private Equity Services