AFTAs 2023: Best new technology introduced over the last 12 months (AI & machine learning)—CompatibL

Project: CompatibL AI

    • WatersTechnology Staff
  • LinkedIn  
  • Save this article
  • Print this page  

Overview

CompatibL AI is designed to manage labor-intensive and time-consuming tasks that would be unfeasible for humans to manage without AI assistance. It can look at every line of source code and every control log message, perform in-depth analysis of the prior documentation and release notes, and integrate all this data in a nuanced way. It can also cross reference the resulting documents with specific lines of source code.

Problems solved  

Historically, vendor-developed trading and risk management platforms and compliance and governance solutions provided a limited degree of automation due to rigid templates. They were also unable to comprehend free-form documents or source code or produce outputs that required integrating multiple disparate sources of information in model governance. CompatibL’s AI offering is innovative in the way it uses large language models (LLMs) to process and integrate information from model specifications, model test results, revision history messages, and regulatory guidelines.

CompatibL’s AI significantly improves the quality of the model governance process, while making the work of analysts less repetitive and more fulfilling. Instead of manually going through hundreds of thousands of lines of model code, revision log messages, and prior documents, the client’s model governance analysts work with CompatibL software and adjust our customized LLM prompts and settings to obtain the desired comprehension focus and generation style that matches their firm’s model governance guidelines.”

Alexander Sokol, founder and executive chairman, CompatibL 

Innovation

CompatibL offers two core models: GPT-4 and Llama 2. As impressive as these LLMs are, they cannot provide specialized comprehension or document generation suitable for model governance. Only customized models enhanced and fine-tuned by CompatibL’s engineers using a curated set of documents and human feedback offer these skills and capabilities. Unique prompt engineering techniques―natural language instructions to the model―are used to obtain the desired results. Embedding techniques developed by CompatibL are used to help the model comprehend large volumes of documents and source code.

Pipeline

The firm’s efforts are aimed at finetuning CompatibL AI and improving the process of documentation generation and the production of model governance documents that can be provided to regulators or banks’ management for model approval if necessary. This complex task requires combining vendor model documentation with many other sources of information across the bank, most of which exist only in natural language form.

Why they won

CompatibL is a name that has, in a relatively short timeframe, become synonymous with innovation. One need look no further than last year’s AFTAs, in which it won the most innovative third-party technology vendor (AI & machine learning) category, and its recent win in the best modeling innovation category in Risk.net’s Markets Technology Awards for examples of its commitment to innovation. Right now, AI is the darling of our industry, although clearly some AI initiatives are more worthy of the tag than others. CompatibL’s use of AI is in the former camp, as illustrated by its win in this year’s AFTAs on the back of its CompatibL AI initiative. Who knows what the CompatibL engineers will come up with next, although if the recent past is anything to go by, it will be both innovative and will address real-world challenges facing its growing list of buy-side and sell-side clients.

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact info@waterstechnology.com or view our subscription options here: http://subscriptions.waterstechnology.com/subscribe

You are currently unable to copy this content. Please contact info@waterstechnology.com to find out more.

Copyright Infopro Digital Limited. All rights reserved.

As outlined in our terms and conditions, https://www.infopro-digital.com/terms-and-conditions/subscriptions/ (point 2.4), printing is limited to a single copy.

If you would like to purchase additional rights please email info@waterstechnology.com

Copyright Infopro Digital Limited. All rights reserved.

You may share this content using our article tools. As outlined in our terms and conditions, https://www.infopro-digital.com/terms-and-conditions/subscriptions/ (clause 2.4), an Authorised User may only make one copy of the materials for their own personal use. You must also comply with the restrictions in clause 2.5.

If you would like to purchase additional rights please email info@waterstechnology.com

More on Awards & Rankings

Most read articles loading...

You need to sign in to use this feature. If you don’t have a WatersTechnology account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here