Egregore Labs was founded in 2017 by Prashant Vijay & Hari Balaji both of whom have spent more than 15 years each working in Tech, Quant and Data roles in Financial Services across New York, London & Asia, primarily with Goldman Sachs.
Our vision is to use AI to transform the use of unstructured data in various Financial Services use cases.
Our customers are primarily Financial Services firms across multiple geographies including a Tier-1 US Bulge Bracket banks & a Top 5 Global PE fund (by AUM)
Financial Services firms receive millions of documents from their customers & counterparties daily; the data within these documents needs to be extracted, understood, processed & stored in business-ready internal formats.
When processing documents that are complex, unstructured, heterogenous or where form & structure are not known in advance / could change without prior notice, traditional automation fails to deliver results. Highly trained subject matter experts familiar with the underlying context become the only viable option.
Romulus - the Solution
Romulus, our flagship product, solves this problem holistically & with high-fidelity through Context-Aware Intelligent Automation.
- Categorize documents into appropriate types, groups or buckets based on content
- Extract the contents of a document (both scanned & machine readable)
- Validate contents against other data sources & for internal consistency
- Normalize relevant content into usable formats
- Persist results into storage / downstream systems as appropriate
- Is powered by a Knowledge Graph that recognizes & utilizes the contextual “DNA” of a document to enhance the quality of its extraction & processing capabilities
- Has been trained on ~100 million diverse documents commonly processed by Financial Services firms both structured (eg. forms) & unstructured (eg. contracts)
This enables Romulus to:
- meaningfully outperform competitors on quality & accuracy when dealing with documents that are complex or vary in format/structure
- implement inter- and intra-document validation to ensure data quality and consistency
- accurately process both structured & unstructured documents
- quickly deliver a high quality solution with limited incremental training & configuration.
Select Romulus Deployments
- Entity Onboarding - Extraction & validation of information from forms, certificates & contracts and auto-population into Entity Master Data Management systems
- Custodial Services - Extraction & record updation from third-party Transfer Agent documents & reconciliation against internal databases
- Middle Office - Segregation of faxes into Trade, FX & Cash instructions & context extraction from fax to create Swift messages / updation of internal systems
- Research Management - Ingestion & tagging of a large corpus of primary research data with assets, authors, concepts, themes etc to support equity research activities
If you'd like further information on these engagements please email us.
Trade Finance Problem Statement
Trade Finance operations require reconciliation of information transmitted via SWIFT messages against documentary evidence that becomes available later. Terms and documentation also need to adhere to the library of international rules. Currently this process is manual requiring heavy involvement from back office personnel. The objective is to use automation to improve the process such that:
- processing time is reduced by 50%
- back-office processes are efficient & streamlined, reducing risks in the process
Romulus for Trade Finance
Romulus is already familiar with a variety of SWIFT Message Formats (including Category 7) & a number of common document types in the Financial Services domain. Romulus can reconcile & validate these supporting documents vs. SWIFT messages.
Romulus delivers the following functionalities:
- Read & decipher MT 700 & subsequent MT 707 messages (as relevant), sent by the Issuing Bank to the Advising Bank
- Read & decipher supporting documents presented by the Beneficiary's bank to the Issuing Bank
- Validate extracted information for internal consistency as well as based on UCP 600 International Rules.
Our web-application supports manual upload of documents & SWIFT message content. We can also deliver the product completely via API through integration with FusionFabric APIs.
We would like to seek the Attijariwafa Bank team's expertise & customization requirements in improving our solution to the next level.