An AI disruption risk assessment template that also scores each company's records

An AI disruption risk assessment for a PE portfolio scores each company on two separate axes: how exposed its revenue and cost base are to AI, and how deep its own operational records run. Plotting both on one heat map shows which exposed companies also hold years of work history that may be worth a data licensing introduction.

What this AI disruption risk assessment measures

The template gives every portfolio company two scores from 1 to 5. Exposure rates how much of the company's revenue, pricing and labor model AI could compress or reshape. Records depth rates how much of the company's own work history sits in systems it controls: tickets, email threads, project files, code reviews, approvals and what happened next.

Most AI risk exercises stop at exposure. Keeping depth as a separate axis matters because the most exposed companies are often those whose daily work looks like the multi-step tasks AI developers now train agents to perform. A services business that resolves thousands of client requests a year is exposed for the same reason its records are interesting: the work is documented, repeatable and judged by outcomes.

The output is a one-page heat map for the investment committee or the annual portfolio review, plus a short list of companies worth a closer conversation.

When should a sponsor run it?

Run it when the portfolio is already being re-underwritten, so the scoring borrows meetings that happen anyway.

Longer holds make it a recurring exercise rather than a one-off. PitchBook reported that the median holding period of US PE-backed companies still in portfolios reached 3.4 years at the end of 2024, the longest in over nine years, with more than 30% held at least five years. A company scored at acquisition can face a very different AI landscape by the time it goes to market.

Moment	Why it fits	Who fills in the sheet
Annual portfolio review	Every company is being re-rated anyway	Operating partner with each deal lead
LP annual meeting preparation	LPs ask how the portfolio is positioned on AI	Investor relations with the operating team
Value creation plan refresh	New levers are being ranked	Portfolio CEO and CFO, reviewed by the sponsor
Add-on diligence	The target changes the platform's exposure mix	Deal team, from management presentation material
12-18 months before exit	The equity story needs an AI answer	Deal lead, CFO and sell-side advisor

The template: one scoring sheet per company

Copy one sheet per portfolio company. Fill it from management conversations and material the sponsor already receives. Nobody needs to open, export or forward company records to complete it.

Company: {company_name}. Sector: {sector}. Year acquired: {year_acquired}. Full-time employees at peak, contractors excluded: {peak_fte}. Part A, exposure (score each 1-5): revenue from tasks AI can already perform {a1}; pricing tied to hours or seats {a2}; share of staff in repeatable desk work {a3}; ease with which customers could substitute an AI tool {a4}. Exposure score, the average: {exposure_score}. Part B, records depth (score each 1-5): years of retained history {b1}; number of core systems holding history {b2}; outcomes attached to the work {b3}; ability to export from current and archived systems {b4}. Depth score, the average: {depth_score}. Part C, licensing gates (yes, no or unknown): the company created the records itself {c1}; client contracts and privacy promises allow licensing {c2}; an owner, CEO, CFO or authorized representative would sponsor it {c3}; the data has not already been licensed for AI training {c4}. Part D, the call: quadrant {quadrant}; next action {next_action}; owner {owner}; review date {review_date}.

How to score exposure

Score each driver from 1 (low) to 5 (high) and average them. Use evidence management can give you in a single conversation.

Exposure driver	Scores 1 when	Scores 5 when	Evidence to ask for
Revenue from automatable tasks	Revenue comes from physical work, regulated sign-off or owned assets	Most revenue comes from drafting, summarizing, routing or answering	Revenue by service line
Pricing model	Priced on outcomes, units or long contracts	Priced per hour or per seat	Rate cards and renewal terms
Labor mix	Most staff do field, plant or licensed work	Most staff do repeatable desk work	Headcount by function
Customer substitution	Switching away means retooling operations	A customer could trial an AI tool within a quarter	Churn reasons and lost-deal notes

How to score records depth

Depth is about the company's own history, not its reporting. The comparison of portfolio monitoring data and operating records explains why monthly KPIs and board packs do not count here.

Depth driver	Scores 1 when	Scores 5 when
Years retained	Under two years, or archives were deleted	5-10+ years, including retired systems
System breadth	Two or three tools	10-15+ systems across email, chat, CRM, finance, support, engineering and operations
Outcomes attached	Work is logged but results are not	Tickets show resolution, deals show won or lost, approvals show granted or refused
Exportability	Nobody knows how to export	IT or the vendor can export each system, including archives

Then answer the Part C gates. The US Copyright Office circular on works made for hire explains that work employees create within the scope of their jobs is generally owned by the employer, while contractor work may not be unless rights were assigned in writing. An unknown answer is fine at this stage; a confident no on any gate parks the company. This is general information, not legal, tax or financial advice. Rights questions belong with the company's own counsel.

Reading the heat map

Plot exposure on the vertical axis and depth on the horizontal one, then sort companies into four quadrants.

Quadrant	What it usually means	Next action
High exposure, deep records	The business model is under pressure, and its work history documents the kind of tasks agents are trained on	Raise data licensing with the CEO this quarter, alongside the AI response plan
High exposure, thin records	Risk without a records asset	Focus on defending revenue; re-check depth after any system consolidation
Low exposure, deep records	A stable business with a possible one-time license	Add licensing to the next value creation plan refresh
Low exposure, thin records	Neither urgent nor licensable today	Re-score at the next annual review

For the top row, the AI value creation playbook for PE operating partners covers the defensive side. A licensing conversation also needs the company to clear the bar on who qualifies: a US business with 50+ full-time employees at peak (contractors excluded), several years of documented operations, rights to license what it holds and an authorized sponsor. Operating teams looking at the wider program can start with referral opportunities for private equity operating partners.

How to adapt the template to your portfolio

Portfolio type	Adjustment
Buy-and-build platform	Score each acquired company separately; archives often sit in the acquired entity's old systems
Software and IT services heavy	Add a depth line for code, pull requests and incident history
Industrials and distribution	Weight exposure toward back-office functions; plant work rarely scores high
Small fund with few companies	Skip the averages and discuss each line in the quarterly review
Holdco or permanent capital	Re-score on a fixed two-year cycle instead of around exit milestones

To work out which relationships beyond the current portfolio might fit, the network opportunity finder helps you think through your wider network.

How often to re-score

Re-score exposure every year, because AI capabilities and customer behavior shift quickly. Re-score depth whenever systems change: a CRM migration, an ERP replacement, a helpdesk switch or an add-on integration can preserve years of history or quietly discard it. Add one line to every new acquisition's 100-day plan: record which systems and archives exist before anything is retired.

What never goes in the sheet

The template is a screening tool, so keep it free of anything sensitive.

No customer names, employee names or record samples.
No exports, screenshots or attachments from company systems.
No estimate of what a license might pay, and no typed reward amounts.
No promises to management about buyers, prices or timing.
No conclusions on rights that only counsel can give.

When a company looks promising, the next step is a conversation with its CEO, not a deeper file review. The guide on introducing portfolio CEOs to an outside program covers how to raise it without pressure.

Next step

Score three companies this month and plot them. If one lands in a deep-records quadrant, register as a partner and make the introduction; SourceX handles qualification and the data inventory directly with the company. For background on how records get reviewed, see what a data audit or assessment is.

Step 1Share your linkSend your personal link to a company you know.
Step 2Company appliesThe company applies itself at /apply.
Step 3Buyer selects and paysThe buyer selects and pays for the data and SourceX receives its fee.
Step 4You get your rewardYour share of SourceX fees becomes payable.

Common questions

Does a high AI disruption score mean a company should license its data?

No. Exposure and records depth are separate questions, and a license only makes sense when the company meets the baseline, holds rights to its records and has an executive who wants to explore it. A high exposure score simply makes the conversation more timely, because the records describe work that AI developers want to learn from. The company decides, and nothing is binding until it signs.

Can the template be completed without requesting company files?

Yes. Every line can be scored from management conversations, headcount reports, revenue by service line and a list of the systems the company uses. Nobody needs to open, export or forward records to fill it in. If a company later explores licensing, it builds its own data inventory directly with SourceX, and redaction rules are agreed before any data work starts.

How can the heat map help answer LP questions about AI disruption?

It shows that every company was scored on the same rubric, where the exposed companies sit and what action each quadrant triggers. That is usually more credible than a narrative about AI strategy. Keep it factual: present scores and planned actions, avoid forecasting license proceeds, and note that licensing is one option for some companies rather than a portfolio-wide response.

Who at a portfolio company is best placed to score records depth?

The CFO or COO, working with whoever runs IT, whether an internal lead or a managed service provider. Between them they know which systems exist, how long history is kept, what was retired and whether exports are possible. The deal lead can sanity-check the result, but the sponsor rarely knows retention details without asking.

What if a company scores high on depth but its records mostly belong to clients?

Then it fails a licensing gate, at least for now. Agencies, outsourcers and some contact centers hold large archives that are really their clients' material, and licensing it would need those clients' consent. Score depth honestly, mark the relevant Part C gate as no, and focus the company's plan elsewhere unless the clients agree.

Free resources

Referral earnings calculator — Hypothetical partner earnings with the per-company cap.
Cash conversion cycle calculator — DIO, DSO, DPO and the cash conversion cycle.
Operational data inventory builder — List systems, record types, years held and owners.
All free tools · MCP resource center

By SourceX Partnerships Team · Published 2026-10-09 · Updated 2026-10-09

Know a US company with valuable proprietary data?

Become a referral partner from anywhere we support, get your link and introduce an owner or authorized decision-maker.

Refer a company →

I own a business

Explore licensing your company's data to AI developers worldwide. Start a short assessment; no uploads needed.

Start an assessment