Watch: Romulus demos its document processing automation software

Romulus specializes in building software products that automate document-heavy operations in financial services. The New Delhi-based Romulus helps customers extract, structure and analyze internal and third-party documents. The company was founded by two former Goldman Sachs executives embedded in technology, trading and derivative design. Romulus’ technology is pre-trained to understand and contextualize more than 200 document types used in financial services. Some of its use cases include: risk management and compliance, new entity onboarding and custodian banking. Unlike optical with character recognition, Romulus’ technology can auto-detect, validate and correct errors based on historical data, contextual clues and validation rules, and offers a “confidence score” after analyzing data that highlights the accuracy of the extraction and analysis. Romulus is part of Fintech Discovery , Bank Automation News ’ sister marketing accelerator program. Watch the video for a short demo of Romulus’ product with co-founders Hari Balaji and Prashant Vijay. Bank Automation Ignite, on April 13-14, is the event for inspiring automation initiatives and investment in financial services. At the virtual event, financial services professionals can discover new use cases and technologies that are accelerating automation in banking. Learn more and register at www.BankAutomationIgnite.com . The following is a transcript generated by AI technology that has been lightly edited but still contains errors. Hari Balaji Hi everyone, I'm Hari Balaji. I'd like to talk to you today about Romulus - a Context-Aware Intelligent Automation Platform built for Financial Services. I founded Romulus with my ex-colleague from Goldman Sachs, Prashant Vijay. Prashant & I have spent much of the past two decades working in Financial Services. Most of that time was spent at Goldman Sachs in Tech, Quant & Data roles. One of the persistent problems we've encountered across geographies, firms and divisions was being able to get information, quickly & correctly out of documents, be it for processing, decision making or just storage. Traditional Automation approaches tend to rely heavily on format, structure & repetition. These work really well when dealing with documents that are standardized in form & structure, and available in large volumes e.g. Tax Forms like W2s. But firms often receive millions of documents every day from thousands of counterparties, vendors or customers. Across many of these situations formats and structure are not consistent. Nor are sample documents available in adequate volume. And in such cases one has no choice but to rely on trained subject matter experts to manually process documents. To solve this problem we need something that truly understands the DNA of a document the way a subject matter expert does and is able to map the contents of a newly received document against the DNA of known documents to contextualize the content. We achieve this by blending Computer Vision & Natural Language Processing with a Knowledge Graph layer and training this ensemble on close to 100 million diverse documents. We further hone this by using our own experience and the network of subject matter experts that we've cultivated over our two decades in the Financial Markets. The result is Romulus a product that delivers Context-Aware Intelligent Automation. Romulus can ingest information across different data sources ranging from email, fax, folders or even an API endpoint. Romulus is able to intelligently contextualize the contents of a document and make nuanced distinctions. For instance it can differentiate between a Certificate of Incorporation and a Certificate of Registration. Romulus can read fields, check boxes and barcodes in a form; and with equal ease Romulus can extract things like payment schedule tables, definitions, clauses and other such information from a Contract-For- Purchase. Romulus can take a set of documents and understand the interlinkages between them to validate and check for inconsistencies, both within a document as well as between documents. Romulus can also reach out into third-party databases and external sources of information to perform validations. Romulus can then standardize all this information into a format and structure of your choosing for export or persistence. Let's now take a quick look at how all of this can come together. Let us start with categorization. In a variety of contexts we see packages of related documents, that need to be processed in conjunction with each other. Romulus can identify documents out-of- the-box and quickly bucket them into appropriate folders. Examples include situations where there is a loan application form or an insurance claim backed by supporting documents or an account opening application. We'll move on to extraction next. The first example of extraction I want to show you is what we call "key-value-pair" extraction from a structured document. This is an investment plan statement. Romulus's objective here is to pull out all the tabular information as well as all the fields and associated values. There is a field - "Holder Account Number" here below which the account number is present. You can see here that Romulus has been able to correctly pick up this association. Right below there are two other fields - "Ticker Symbol" and "CUSIP". The "Ticker Symbol" here is "HYT" and the "CUSIP" is this alphanumeric number. Traditional extraction tools will rely overly on proximity and using the previous example end up mapping "Ticker Symbol" - the field against "CUSIP" its value. Romulus is smart enough to figure out that this is not the case and to get it right. Romulus is also clever enough to be able to extract all the tables in this document even though there are no clear boundaries or boxes around them. Next I'm going to show you a highly unstructured document - a Credit Agreement. Romulus is set up here to work with a "Human-in-the-loop". Every piece of extracted information, which in this case includes things like clauses, and definitions, is placed in a text-box allowing for manual overrides. Romulus also provides information regarding its confidence in the quality of extraction which serves as cues to the human reviewer. Let's look at "Governing Law" for instance. "Governing Law" here is set up to extract an entire clause. And when I click in, the Document self- navigates to tell me where that particular clause is. After reading this entire section I could choose to simply update with the words "New York Law", hit Save and add a comment, and this will be persisted in the system. Finally I'd like to show you an end-to-end deployment of Romulus for a Trade Finance use case. Each row here represents a letter-of -credit across different stages of processing. I am going to click review to view a particular example each Letter of Credit here consists of its terms (or amendments to those terms), which are presented in SWIFT message format. Subsequently supporting documents such as Bills of Lading, Commercial Invoice, or Packing Slip are submitted. Once this Documentary Evidence Submission is complete, Romulus is able to perform a variety of validations and rule checks to surface issues. For instance, in this particular submission, there is an invalid UCP code and further a check of the transporting vessel against the OFAC Sanctions List suggests that this could be a Sanctioned Vessel. We went through several examples of Romulus's capabilities across Categorization, Extraction and Validation and how Romulus can automate an entire workflow end-to-end. I also want to mention that Romulus is able to read from, write to, and integrate with, a variety of third-party software and able to automate both Autonomously as well as with a Human-in-the-loop. We are ISO-27001 certified and come with Enterprise-grade privacy & security. Romulus can be deployed on-prem in the cloud or both. I'd like to thank Bank Automation News for giving us the chance to present to you. To know more about Romulus visit www.romulus.co or email us at sales@romulus.co. Thank you!

With $220M Series F, WorkFusion looks to expand tech

Connect

Welcome Back!

Retrieve your password