Linklaters

The UK bank ring-fencing reforms, coming into force in January 2019, required banks holding over £25bn in assets to establish separate retail banking divisions — creating an unprecedented volume of structured due diligence work for Magic Circle firms like Linklaters. The firm's banking practice needed to systematically analyse large bodies of legal documents to classify institutions and assess contractual obligations under the new legislation. Existing NLP tools from RAVN Systems and Kira Systems, while capable, couldn't provide the granular flexibility and workflow control Linklaters required across both its transactional and advisory departments. Manual analysis at the scale demanded by ring-fencing work was not viable.

Banking partner and AI working group head Edward Chan led Linklaters' collaboration with London tech startup Eigen Technologies — whose co-founder Lewis Liu was a former senior strategic adviser at the firm — to build Nakhoda, a proprietary NLP-backed data extraction and analysis tool. The relationship was grounded in shared institutional knowledge: Eigen understood Linklaters' workflows from the inside. Nakhoda was deployed alongside existing tools in a deliberate 'toolkit approach', sitting next to RAVN Systems and Kira Systems rather than replacing them. This multi-vendor architecture reflected the firm's view that no single NLP solution could cover the full spectrum of transactional and advisory use cases. A companion tool, LinkRFI, was also built to classify institutions under ring-fencing definitions at scale, drawing on external registers for credit institutions and insurers.

Nakhoda moved into active testing across both due diligence and internal know-how workflows, with discussions underway for external client deployment — a meaningful step for a firm-built tool. LinkRFI, the related classification tool, was adopted by Lloyds Banking Group and Royal Bank of Scotland to screen millions of institution names in a fraction of the time previously required for manual memo-based review. Key outcomes included:

• Dramatic time reduction in classifying financial institutions under ring-fencing definitions, a task previously requiring individual legal memos
• Cross-departmental adoption spanning both transactional and advisory practices
• Client-facing deployment of IT-built tooling, described by Chan as 'deploying IT on the front line'

• Build for flexibility, not feature parity — Linklaters built Nakhoda specifically because off-the-shelf tools couldn't match its workflow requirements, not because they were inferior products overall.
• Leverage institutional relationships with vendors — the Eigen partnership worked because the founder understood Linklaters from the inside; prior relationship accelerated trust and fit.
• A toolkit approach reduces single-vendor risk — no NLP tool covers every legal use case; running multiple tools in parallel is a more resilient architecture than seeking one solution.
• Lawyer-led AI initiatives outperform IT-only efforts — banking partners co-leading the project ensured the tool addressed real practice needs rather than theoretical ones.