Among the most pressing operational challenges facing the banking sector today is the need to separate retail and investment banking. The process of reverse engineering sophisticated global financial institutions is complex. Achieving separation successfully and efficiently is crucial. Working with a number of the world's largest banks Pinsent Masons has deployed a unique blend of people, process and technology, including our proprietary AI software, to deliver just that.

The challenge

In the wake of the global financial crisis, the UK parliament passed the Financial Services (Banking Reform) Act 2013, which forces is the separation of retail and investment banking operations by 2019.

This process requires the novation of millions of legal documents from one entity to another through court approved Part VII transfers. This process can include everything from loans and securities to loan guarantees, indemnities, letters of credit and other documents.

Typically novation would require consent from the counterparty. It would, however, be impractical and unnecessary to approach counterparties on all agreements on a project of this scale. So the banks can apply to the courts for a Part VII transfer which overrides the requirement for customer consent.

The challenge for the banks is to:

  • identify which documents should legitimately be included in a Part VII request
  • ensure the court has an appropriate level of detail regarding the portfolio to make an informed decision
No other firm has invested so heavily, over such a sustained period, in developing its own AI platform.

The solution

Pinsent Masons has developed proprietary technology, TermFrame ERA, to apply machine learning to this unique instance of diligence.

A key benefit of developing our own platform is that it can be quickly configured to meet different client challenges.

The platform has been contextualised to support ringfencing programmes. New functionality was developed including:

  • an automated data extraction component, which ingests and clusters huge volumes of documents, and extracts key provisions;
  • a tool allowing a bank to 'cluster' all documentation fundamentally derived from one precedent together and highlight the variations. This adds another layer of transparency over the portfolio;
  • review, analysis and reporting tools bespoke to client requirements.

No other firm has invested so heavily, over such a sustained period, in developing its own AI platform.

It was a strategic decision to incubate the talent and the technology in-house. The benefits of that approach are now becoming clear. The firm has a technology platform flexible enough to incorporate the technical insight of the legal team and respond fast.

The result

  • For one client alone we have reviewed over 23,000 documents using TermFrame ERA
  • Projects have ranged from Detailed Review programme to a High-Level Diligence review, completed using data extraction only. On the Detailed Review programme, 1879 documents were reviewed in just 22 days at an average run-rate of 77 documents per day.
  • On the High Level Diligence review, over 12,000 documents were reviewed in 30 working days, yielding an average run-rate of 405 documents per day.

No other law firm could rival these outcomes. Those continuing to rely on substantial manual review would struggle with the volumes, and costs would spiral out of control. We calculate that the average cost of per document in our programmes has been between around 97% lower when compared to manual review - and 63% lower for more complex documents.

Other technologies are available and are being used to support similar processes. However, our clients tell us ours is the only team hitting its deadlines.

Further, the technology is being adapted for use on remediation projects, securitisation diligence and other balance sheet management transactions. There will be many further applications in other sectors.

We are always happy to demo our tech to clients and potential clients – just get in touch.

Out-Law / Your daily need to know

We are processing your request. \n Thank you for your patience. An error occurred. This could be due to inactivity on the page - please try again.