AI readiness assessment for portfolio companies: what to check and what it tells you

Why assess AI readiness before funding AI projects

Most portfolio AI projects that disappoint fail on inputs, not ideas: the data is scattered, nobody owns the workflow, or a contract forbids the intended use. A readiness assessment finds those blockers before money is spent and gives the board one consistent view of where each company stands.

It also does a second job. The systems inventory at the center of the assessment, listing every tool, how many years of history it holds and who can export from it, is the same evidence that shows whether a company holds records AI developers would license. Mature companies often run 10-15+ systems and keep 5-10+ years of history. One inventory therefore informs two decisions: what to build with AI, and what the company already owns that AI developers want. The AI value creation playbook explains both tracks.

How to run the assessment in two weeks

Keep it light enough that the CEO agrees without a fight.

1. Agree scope with the CEO and CFO. Explain that the output is a readiness score and a short action list, not an audit.
2. Request five documents: the list of software subscriptions, the data retention policy, the customer-facing privacy policy and terms, a standard customer contract, and the IT or MSP service agreement.
3. Interview four people for 30 minutes each: the CFO, the head of operations, whoever runs IT, and one frontline manager from the highest-volume workflow.
4. Build the systems inventory: each system, what it records, the oldest record still available and whether a full export is possible.
5. Work through the checklist below and score each area.
6. Present the results at the next board or operating review with two or three actions, not twenty.

The data inventory builder helps a company list its systems and records in one place.

The AI readiness checklist

Data

• Core workflow data (orders, tickets, projects, invoices) lives in systems, not in spreadsheets on personal drives.
• Key fields are filled in consistently enough to report on without manual cleanup.
• Outcomes are recorded: why a deal was lost, how a ticket was resolved, whether a project finished on budget.
• Someone can name where the authoritative copy of customer, product and employee data sits.
• History from replaced systems was exported and kept, not deleted at migration.

Systems

• A current list exists of every business system, its owner and its renewal date.
• The main systems can export data in bulk, and someone has done so recently.
• Systems that share customers or projects are linked by common IDs.
• Retention settings for email, Slack or Teams are known and written down.
• The MSP or IT contract covers data access and export requests.

People

• A named executive owns AI decisions and the AI budget.
• Each candidate workflow has a manager willing to redesign it.
• Staff have a written policy on which AI tools they may use and with what data.
• Someone in-house can run exports, scripts or reports without calling a vendor.

Governance

• The privacy policy and customer terms have been read for promises about how customer data is used.
• Customer contracts have been checked for confidentiality and data-use clauses.
• Employee and contractor agreements show who owns work product.
• Any call recordings were made with the required notices.
• Health, financial or consumer personal data has been identified and is handled under the applicable rules.

Two governance items deserve extra care. FTC staff have stated that a company's promises not to use customer data for undisclosed purposes, such as training or updating models, are enforceable whether they appear in privacy policies, terms of service or marketing, so read what the company promised before any AI use or license. On ownership, the Copyright Office's Circular 30 explains that work an employee prepares within the scope of employment is a work made for hire owned by the employer, while commissioned work qualifies only in listed categories with a signed written agreement, so material produced by contractors may need a written assignment. This is general information, not legal, tax or financial advice. Confirm with the company's own counsel before acting.

Licensing read-out

These items mirror the who qualifies baseline.

• The company is US-based with 50+ full-time employees at peak (contractors excluded).
• It has several years of documented operations, with that history still retrievable.
• Its records span many systems: email and chat, CRM, finance, support, engineering or operations.
• The records are the company's own, not mainly a client's material.
• The data is not mainly consumer personal information or protected health information.
• The same data has not already been licensed for AI training.
• An owner, CEO, CFO or authorized representative could approve a license.

How to use the results

Score each area as ready, partial or blocked, then read the combination rather than the total.

A pass on the licensing read-out justifies a conversation, not a conclusion. The questions to ask a portfolio CEO about AI help frame it, and the explainer on whether a PE portfolio company can license its data covers what the CEO will want to know first.

What the operating partner does with a licensing pass

Raise it with the CEO and, if there is interest, introduce the company to SourceX. Do not forward the inventory, screenshots or samples. The company works with SourceX directly on qualification, its own detailed inventory, the rights review and any redaction rules, and nothing is delivered without an executed agreement and the company's authorization.

Partners earn 25% of the eligible platform fees SourceX actually collects from the referred company's licensing deals, capped at $100,000 per referred company. The reward is paid only after the buyer pays and SourceX receives its fee; an introduction, meeting or signed agreement alone does not trigger payment, and no reward is guaranteed. The operating partner program page explains how the referral works for your role.

Red flags in a readiness review

• Retention settings delete email or chat after a short period, and nobody knew.
• A system migration is scheduled and the old platform will be switched off without a full export.
• The company's records are mainly a client's work, as at many agencies and outsourcers.
• The privacy policy told customers their data would never be shared or used for other purposes.
• Nobody inside the company can export data without a paid vendor project.
• The AI budget assumes savings that no one has measured.

Next step

Run the checklist at one portfolio company before the next budget cycle and let the result decide where AI spending goes. Read how AI affects EBITDA to model what the buy side is worth. Where the licensing read-out passes, register as a partner and introduce the company, or point the CEO to sourcex.si/apply through your referral link.