AI Readiness in PE Portfolios: Why Pilots Rarely Reach the P&L

Portfolio companies keep launching AI pilots that never appear in a management account. Experimentation with AI is now widespread across portfolio companies, but the translation of that activity into sustained, P&L-level impact remains uneven. McKinsey Most operating partners track adoption speed instead of readiness, and the two measure different things. This article outlines the four-pillar readiness framework operating partners can use to audit portfolio companies before the next board cycle.

AI readiness is a portfolio company’s capacity to convert an AI pilot into a repeatable, measurable change in cost or revenue, not just a working demo. This distinction matters because sponsors now price AI capability into deal terms, not just deal diligence.

A portfolio company is AI-ready when its finance data, workflows and staff can absorb an AI tool and turn it into a tracked line-item change within one reporting cycle.

• Require every portfolio company to name a specific P&L line item before approving further AI pilot budget.
• Run a data and workflow audit ahead of any new AI investment. Fragmented finance or ops data is the leading reason pilots stall.
• Score each portfolio company against a four-pillar readiness scale (data, workflow, skills, measurement) before the next board cycle.
• Treat AI capability as a re-underwriting input when reviewing add-on acquisitions or exit positioning, not a side project.
• Assign one operating partner per portfolio company to own AI-to-EBITDA tracking, separate from the technology vendor relationship.

AI readiness for a portfolio company is the degree to which its data infrastructure, workflows and staff can absorb an AI tool and convert it into a repeatable operational change. A pilot running on a laptop, disconnected from the general ledger or the ERP, has not solved a readiness gap. It is a demo that will not survive quarter-end reporting.

Operating partners who treat readiness as a technology question consistently see pilots stall at the proof-of-concept stage. The failure is not the model. It is the absence of a workflow built to receive the model’s output and turn it into a booked cost reduction or revenue gain.

For portfolio companies attempting to close this gap on accounts payable (AP), reconciliation or reporting workflows, tkxel’s AI Workflow Automation approach treats the finance workflow itself as the unit of change, not the model in isolation.

AI pilots stall before reaching the P&L because most are built to prove a capability, not to replace a cost or generate revenue inside an existing workflow. A pilot proves a model can summarize contracts or draft customer replies. It rarely proves the summarization step eliminates a full-time role, or that the reply tool reduces support headcount.

Experimentation with AI is now widespread across portfolio companies, but the translation of that activity into sustained, P&L-level impact remains uneven. McKinsey That gap between demonstrated capability and captured value is the single biggest reason boards see AI activity without seeing AI results.

Three structural problems drive this gap:

• Fragmented data across the finance stack, so the AI tool operates on a partial or stale dataset instead of the full transactional record.
• Pilots run in isolation from the workflow they are meant to change, so the human process continues in parallel and no cost actually shifts.
• No line-item ownership, so no single leader is accountable for tying the pilot’s output to a specific cost reduction or revenue increase.

AI readiness has become a valuation question because sponsors now factor AI capability directly into how they price and manage portfolio companies. AI is increasingly forcing sponsors to re-underwrite their portfolios, not just assess new deals differently. McKinsey That shift changes what an operating partner needs to know about every company in the book, including add-on targets under diligence.

It is no longer sufficient to ask whether a company has an AI strategy. The relevant question is whether its finance function, customer operations or supply chain can absorb an AI-driven change and hold the resulting margin improvement through the next reporting cycle, and through an eventual exit process.

Thoughtworks’ 2025 State of Digital and AI Readiness report found that only a small fraction, 17%, are truly leading the market in digital and AI readiness Thoughtworks, a figure covering organizations globally across industries, not portfolio companies specifically. Applied to a typical book of ten to fifteen companies, that ratio suggests most operating partners manage portfolios where a majority of assets have not cleared the readiness bar sponsors now underwrite against. One readiness leader in a portfolio does not offset a majority still running disconnected pilots. That gap is exactly what shows up during add-on diligence, when a target company’s AI claims get tested against its actual finance data structure.

The core components of AI readiness are data governance, workflow integration, organizational capability and measurement infrastructure. Each pillar gates the next, so weakness in one caps the return available from the others.

Data governance requires the company’s finance, ops and customer data in a single governed source, owned by a named function. Spreadsheets and disconnected systems inherited by AI tools replicate the fragmentation humans already work around.

Workflow integration means the AI output feeds directly into the process it is meant to change, such as the AP queue or the customer service ticketing system. A tool that sits in a separate dashboard someone has to check manually has not integrated with anything.

Organizational capability means the staff running the affected workflow have retrained around the new process. A change management plan specifies who approves exceptions and who owns the transition period.

Measurement infrastructure means the company defined, before the pilot started, which P&L line item the initiative targets and how movement gets tracked month over month.

A digital transformation readiness assessment across these four components gives an operating partner a defensible answer to the question every board eventually asks: is this initiative working, or is it just running.

To run an AI readiness assessment across a portfolio, an operating partner works through four sequential steps rather than scoring technology maturity alone.

1. Audit the data and workflow foundation at each company, identifying where finance and ops data sits fragmented across disconnected systems.
2. Evaluate how deeply any existing AI tool integrates with the actual finance or operations workflow, not just whether the tool exists.
3. Assess staff readiness and change management capacity, including who owns the transition and who approves exceptions during rollout.
4. Tie every active or proposed pilot to one specific P&L line item, with a defined measurement window and a named accountable owner.

This is an ai readiness assessment applied at portfolio scale, not at the level of a single tool purchase. Running it before approving new AI budget prevents well-funded pilots that never connect to a reporting line.

The table below states how readiness signals differ between a pilot-phase company and a production-phase company, based on the structural patterns described above.

Sponsors now build re-underwriting into portfolio reviews specifically because AI is increasingly forcing sponsors to re-underwrite their portfolios, not just assess new deals differently. McKinsey A company stuck in the pilot-phase column has no defensible answer when that review happens.

tkxel, a B2B software engineering and AI services company, builds intelligent products and deploys agentic systems to production for mid-market and growth-stage businesses. The methodology centers on connecting AI initiatives directly to existing finance and operations workflows, rather than deploying standalone tools that require a second system of record.

tkxel has supported organizations across SaaS, healthcare, fintech and enterprise platforms in moving AI programs from pilot to measurable operational use, including workflow automation inside finance functions such as accounts payable and reconciliation. The focus stays on the question every operating partner asks in a board review: what changed on the P&L, and can it be tracked next quarter.

The readiness gap covered in this article is a workflow, data and accountability gap, not a technology gap. It shows up the same way in every portfolio company that cannot connect an AI pilot to a reporting line.

Operating partners weighing where their own portfolio stands can start with tkxel’s five-point framework in Is Your Business Ready for AI Automation, which maps directly onto the four pillars covered here, or review how sequencing affects EBITDA outcomes across a multi-company book in The PE Technology Playbook. The next step is not another pilot. It is an honest audit of which readiness pillar is actually missing.