Compliance Data Services
Menu

AI readiness evaluation

Get your AI readiness score

Before you rely on AI for applicability, R2 tests whether your existing taxonomy gives it the context to decide, and scores it.

How the evaluation works

From your taxonomy to your score

You don’t need another platform. R2 works from your existing data. Most organizations export requirement records from their GRC platform.

  1. 01Structure test

    R2 tests whether each field in your requirement records, such as the requirement description, holds its information in a form AI can use reliably.

  2. 02Applicability test

    R2 tests whether each requirement’s record, read in full, gives AI the context to decide which of the obligations it cites apply to your firm.

  3. 03AI Readiness Score

    The two tests score your firm’s regulatory taxonomy, producing your AI Readiness Score.

  4. 04Suggested modifications

    R2 presents your AI Readiness Score and suggested modifications to your team.*

*Fixed- and variable-price .

*Updates are handled under a separate engagement.

Example report

AI Readiness Scorecard

The evaluation tests whether your taxonomy gives AI enough context to decide applicability accurately and reliably, the decision on which regulatory change triage, impact assessment, risk assessment and testing scope depend. Your taxonomy is the set of requirement records and their fields exported from your GRC platform: each row is a requirement record, each column a field, such as Topic, Citation or Description.

The structure test checks each field’s value for a form AI can use reliably. The applicability test considers each complete record and checks whether AI reaches the applicability decisions your experts have approved for the obligations it cites.

AI readiness evaluation

Example report

AI Readiness Scorecard

Prepared by R2 Compliance Data Services

Assessed October 14, 2026 · Evaluation set v1 · AI model as set in the engagement

Overall score
38.5
Applicability
32.0
Structure
51.5

Scope: 100 requirements · 13 fields

Applicability, by requirement

Whether an obligation applies depends on the businesses, legal entities, locations and products it reaches. This test checks whether each requirement’s record, read in full, gives AI enough context to decide applicability accurately and reliably: whether each obligation it cites applies, and to which parts of the business. That context is spread across the record’s fields, so the test reads them together. The pass rate is the percentage of the 100 requirements whose record passes. Select the row to see the reasons for failure and an example.

MeasureResultFinding
32.0%

Structure, by field

Whether each field in the requirement record, such as the requirement description, has what AI needs, in a form it can use reliably. Each pass rate is the percentage of the 100 records whose value in that field passes. This example shows 13 of the fields R2 evaluates. Select a field to see the reasons for failure.

FieldResultFinding
56.0%
55.0%
65.0%
30.0%
62.0%
39.0%
40.0%
24.0%
60.0%
70.0%
49.0%
62.0%
57.0%

How the scores are calculated: A requirement passes applicability when AI, working from its record, reaches the decision your experts approved; the applicability score is the percentage of the 100 requirement records that pass. The structure score is the average pass rate of the 13 fields shown. The overall score weights applicability twice as much as structure; it is a composite of the two tests, not a measure of AI accuracy. Each failing requirement is counted once, under its primary reason. The checks follow the data standards agreed with you before the evaluation; this example uses common ones.

Questions

What to expect

How common are these problems?
Common. In the taxonomies R2 has worked with, citation formatting error rates have run as high as 60%. In the case study on the home page, the taxonomy’s requirement topics had no scope notes and its requirement records did not state who must act or what they must do, and AI reached wrong applicability decisions as a result.
What do we need to provide?
An export of your requirement records from your GRC platform. They usually already cite their regulatory sources, so you choose which sources to evaluate.
How long does an evaluation take?
It depends on the number and size of the sources you choose. R2’s proposal sets out the timeline before work starts.
Who makes the changes?
Making the changes is a separate engagement. After the evaluation, R2 can update your taxonomy on your behalf or alongside your team, or your team can make the changes itself.
How does the evaluation decide that a requirement passes applicability?
Your experts approve a set of applicability decisions for your taxonomy. A requirement passes when AI, working from its record, reaches the approved decision. No second model judges the result. Records pass or fail as units.
Which AI does the evaluation test?
AI models: During the initial engagement and project scoping, the team will determine which generative AI model should be used for the evaluation. *Model selection might affect the cost of the evaluation. Your taxonomy and the decisions your experts approve are yours: they work with whichever model you choose, and outlast it.
What happens to our taxonomy?
R2 uses it only for your engagement and keeps it for the length of the engagement: the evaluation, and the ongoing services if you continue. When the engagement ends, R2 deletes it.
How is the score calculated?
The evaluation applies two tests, and each check passes or fails. Applicability tests whether each requirement’s record, read in full, gives AI enough context to decide accurately and reliably whether each cited obligation applies, and to which parts of the business; its score is the percentage of requirement records that pass. Structure tests whether each field in the record holds the base information AI needs; its score is the average pass rate of the fields. The overall score weights applicability twice as much as structure, because applicability is the decision the other work depends on.
Does the score assess our compliance program?
No. The score measures how well your taxonomy supports AI in determining applicability and in the internal flows that follow it. It does not assess whether your controls are effective.
What evidence does the evaluation give our AI governance, model risk or internal audit teams?
Two records. Before you rely on AI: whether the data it works from is complete enough to decide applicability, and what to change. After, each time the taxonomy, the regulations, the model or its prompts change: whether AI still decides as your experts would, or has drifted.

Start with an AI readiness evaluation

Does your regulatory taxonomy give AI enough to decide?

Request an evaluation

Tell us what you would like assessed. We’ll reply to discuss scope, timing and next steps. Not sure yet? See how the evaluation works first.

Fixed price covers a subset of the FCA Handbook or the US Code. Variable price covers all other regulatory sources.

If you have one from an event or an offer.

We’ll only use these details to reply to you. Privacy Policy (opens in a new tab)