Financial institutions operating across multiple jurisdictions face a growing challenge in keeping up with regulatory change. To support them, we’ve developed a hybrid legal intelligence approach that combines generative AI with expert legal review. This allows us to build and maintain structured inventories of laws, rules and regulations, reducing manual effort while ensuring legal accuracy through a human-in-the-loop process.
The solution has been used to help clients map and track regulatory obligations across the UK, EU and Middle East. By breaking down complex legal texts into clear, actionable requirements and tagging them by priority, we’ve enabled teams to manage compliance more efficiently, support internal controls mapping and prepare for future automation. It’s a practical, scalable way to stay ahead of regulatory change.
The challenge
Large regulated organisations are under constant pressure to keep up with evolving rules, policies and standards. Manually tracking and mapping these changes to internal frameworks is slow, resource-intensive and prone to error.
The key challenges we’ve addressed include:
- Volume: Processing large volumes of regulatory content from varied formats like HTML, PDFs and scanned documents.
- Complexity: Distinguishing between meaningful legal changes and minor editorial edits across multiple versions.
- Fragmentation: Aligning overlapping or inconsistent data from different sources and jurisdictions.
Our approach
We developed a hybrid legal intelligence approach combining generative AI with expert legal review.
Phase 1: scoping, tiering and decomposition
In the first phase, the system ingests regulatory materials using a bespoke pipeline that scrapes and processes static documents into a structured XML format. This allows for consistent handling of content from a wide range of sources and formats.
Once ingested, the content can be structured into any template or format, allowing inventory to be tailored to specific compliance or regulatory requirements.
Once ingested, a large language model (LLM) analyses the inventory to identify, summarise and tag individual obligations. The LLM also performs an initial “decomposition” of the legal text, breaking it down into discrete provisions and mapping them to relevant themes. Semantic search can then be used to generate an initial link to related rules and guidance, allowing regulations to be assessed in conjunction with the related context.
Phase 2: legal adequacy review, briefings and updates
In the second phase, legal experts assess whether the controls, standards and policies used by the institution are sufficient to ensure compliance with each of the obligations identified in the inventory. This is a complex analysis and the LLM can be deployed to speed up and supercharge the adequacy review by compiling controls, standards and policies into a single source and undertaking a thematic review relating to each obligation. This helps the lawyers identify the relevant controls faster so they can spend their time doing the detailed legal analysis, rather than compiling source material and locating content.
The result is a high-quality, legally robust obligations register that can be used for compliance mapping, horizon scanning and future automation initiatives.
The LLM is also used to prepare stakeholder briefings and updates quickly and efficiently. Using different prompts, the LLM can prepare summaries of key regulations and highlight critical areas of compliance. The LLM can also ingest “old” and “new” regulation text and summarise the changes and impact on the business or product, helping legal teams quickly understand new developments.
At all stages, all LLM outputs are checked by Simmons lawyers. Prompts are engineered and iterated with subject matter experts.
The outcome
A high-quality, legally robust obligations register used for compliance mapping, horizon scanning, and future automation. The solution has helped clients map and track regulatory obligations across the UK, EU, and Middle East. All LLM outputs are checked by our lawyers.
- UK · EU · Middle East
- Human-in-the-loop
- 100+ jurisdictions
- Weekly automated updates
We’re not consultants guessing at your pain points.
We're lawyers, solutions engineers and data scientists who have lived them. Tell us where you are on your journey and we'll help you chart the path forward.
_11zon.jpg?crop=300,495&format=webply&auto=webp)
.jpg?crop=300,495&format=webply&auto=webp)
_11zon.jpg?crop=300,495&format=webply&auto=webp)
_11zon.jpg?crop=300,495&format=webply&auto=webp)
_11zon.jpg?crop=300,495&format=webply&auto=webp)
.jpg?crop=300,495&format=webply&auto=webp)