Tellimer releases version 2.0 of Parsel, ita artificial intelligence (AI) powered data extraction tool.
Parsel is an intelligent document processing (IDP) tool, curated by the financial technology, data and information provider’s data science and engineering team to enable financial and capital-market participants to better process credit, sovereign, company and other financial information.
The tool is capable of analysing bond documentation, term sheets, pricing supplements, derivatives contracts, and financial reports in an effort to automate the identification, extraction and digitisation of target information
At a technological level, the tool applies advanced computer vision, statistical analysis and natural language processing to extract and digitise specific pieces of text, tables, terms and forms from financial and capital markets documents.
The upgrade now includes different algorithm configurations for different data formats and use cases and incorporates a flexible API with which in-house teams can configure bespoke solutions for clients.
V2.0 seeks to enable businesses to process large volumes of documents at scale, including term sheets, pricing supplements, OTC derivatives confirmations, financial reports, invoices, purchase orders, business forms, legal documentation, macroeconomic datasets and procurement data.
A test of the upgraded tool demonstrated significantly greater accuracy and fewer output errors than any other IDP or data extraction tool currently on the market; as per the company’s official statement.
As the role of data management in operational efficiency and market advantage continues to grow, Parsel estimates that the digital transformation market is now worth $1trillion, of which data and table extraction accounts for $1billion, growing at a 13 per cent annual CAGR.
The financial and investment industry is naturally documentation-heavy with multiple conventions in naming comparable values, concepts or metrics; elements the tool seeks to solve.
Commenting on the 18-month process of launching the upgrade, Daniel Vliegenthart, the company’s tech lead, describes how the update pioneers data accuracy, citing testing of the tool as an example.
“With this new release, accuracy has improved still further when tested against our broadest and most challenging set of test data,” he comments.
Vliegenthart says the release “cements Parsel’s superiority in the field of table and form field extraction compared to our competitors.”
Adding to these comments, the company’s CEO, Duncan Wales, describes the birth of its homegrown solution.
“We built Parsel for ourselves to accelerate our own analysis of companies and credits, and decided it was so good we should share it with our clients,” he reveals.
Wales admits that the company realises first-hand how hard it is to locate and transpose terms, clauses, data, metrics, financial reporting and forecasts from financial documents.
“Having built it to financial levels of accuracy we also anticipate it will be a powerful tool for enterprise clients in a number of industries, as the need for secure and accurate digitisation accelerates,” he adds.
The upgraded tool is available via the company’s web application on Parsel.ai or can be integrated directly into existing business workflows via the customisable API.