For compliance and risk teams at banks and heavily regulated industries
Your taxonomy was built for people.
R2 prepares it for AI.
Your GRC platform stores what people need, not what AI needs. R2 updates the mappings you already have and provides a regulatory library to match, so AI can triage applicability and gaps at the obligation level, linked to your policies, procedures, controls, and tests.
MappingHow a regulatory mapping works
Mappings
Organized by internal categories
What banks track today
- Risk ratings
- Control owner
- Policy status
- Impacted LOBs
- Mapping notes
Not recorded
- Definitions
- Context
- Verified source links
Regulatory requirements
Needs
- Citations
- Obligations
- Summaries
Internal compliance processes and documents
Linked by hand
- Policies
- Procedures
- Controls
- Tests
Links are maintained by hand. AI has to guess what each category means and why each link exists.
How R2 enhances
- New mappings for AI
- Regulatory library
AI triages applicability and gaps at the obligation level, down to policies, procedures, controls, and tests
01 — The problem with regulatory data
Can large language models (LLMs) process Regulatory Structure correctly?
Regulatory structures differ across jurisdictions, and inconsistencies in document structures wreak havoc on using AI for data conversion.
QuebecPrivacy act P-39.1p. 9
Structural issue
The page carries a second, invisible copy of the citation text.
What breaks
The citation resolves to a different section: s. 110 becomes s. 1101.
Expected structure
- 2021, c. 25
- s. 110
Broken structure
- 2021, c. 25
- s. 1101A different section
02 — The problem with client data
Does your taxonomy increase or decrease errors when using AI?
R2 has found that system of record (GRC) systems are out of sync. Your team fills the gaps from experience. AI has to guess.
What guessing looks like
There’s no field for what the AI needs.
Obligation
An MCD article 3(1)(b) creditor must comply with CONC 2.4 (credit references: conduct of business: lenders and owners).
Linked by AI using the current mapping filematched on “credit reference”
Topic the AI chose
FCRA requirements for permissible purposes and credit information handling in commercial real estate lending
AI justification
CONC 2.4 addresses credit reference conduct. This topic covers FCRA requirements for permissible purposes and credit information handling in commercial real estate lending — the same subject matter applied to a different lending context.
What went wrong
It linked a UK consumer credit rule to a US topic on commercial real estate lending. It noted the different context and linked them anyway.
What the data didn’t say
That the topic’s jurisdiction and lending context are limits, not just description.
In each case, the AI justification made incorrect assumptions from the client’s internal mapping data.
Obligations from the FCA Handbook · Not client data
03 — How R2 works
Map at the level of detail the work needs, not the level your team can maintain.
Three parts. R2 does the pre-work on the mappings you already have and provides a regulatory library built to the same level of detail. Together, they enable applicability and gap analysis at the obligation level.
01
New mappings for AI
R2 updates your existing taxonomy mappings to include the structure and data a large language model requires to operate with precision. It checks each existing mapping against the source and recommends enhancements.
- Each category carries the scope, inclusions, exclusions, and contrasts AI needs.
- Each link records its source and reason, for your approval.
Like your existing taxonomy, the AI mappings have to align with the way your company perceives its risk.
02
Regulatory library
R2 turns each source into structured requirements with precise citations.
- One consistent structure, with citations at any level of detail, linked to the text.
03, enabled by parts 01 and 02
Applicability and gap analysis at the obligation level
Your new mappings and the regulatory library work together, so AI can assess applicability and find gaps for each obligation, at scale.
- Each obligation triaged for applicability, with a justification tied to its source.
- Gaps flagged where an obligation has no linked control, policy, procedure, or test.
Finer detail, same team. Your regulatory library and AI triage applicability and gaps together, so your team can manage at a much finer level of detail without adding headcount.
Regulatory change, in the process. With your data at the right level of detail, a regulatory change runs through the same applicability and gap analysis, down to the controls, policies, procedures, and tests it affects.
Your team adds judgment.
AI adds scale.
Your experts keep responsibility for interpretation and approval.
Start with an assessment
Are your regulatory mappings ready for AI automation?
Start with one internal taxonomy. R2 shows where AI would have to guess during applicability and gap analysis, and which mappings to enhance first.
What you receive
- Findings on missing definitions and unsupported relationships
- Supporting citations for each finding
- Recommended mapping enhancements for your approval