Working for Faster Innovation Using EBA with TP ICAP

Parameta is the data and analytics division of TP ICAP—a provider of financial markets infrastructure and data to investment and regional banks, hedge funds, asset managers, corporates, and trading houses. Parameta provides market data, indices, and analytics solutions that empower its clients to drive smarter trading, manage risk effectively, and stay ahead of regulatory demands.

 

When Parameta wanted to build customer-focused solutions using generative AI, it turned to AWS because it sought the highest-performing tools for transforming any organization’s workflows. Parameta’s data science and data engineering teams worked on a prototyping engagement alongside AWS solutions architects and product managers. The engagement gave Parameta’s teams hands-on experience on AWS technologies and ways of working, such as the “Working Backwards” principle. The team developed a regulatory assessment tool from concept to a production-ready solution in the 3-week engagement. “We believed that we could build a scalable generative AI solution using this approach,” says Martin O’Gorman, head of data science at Parameta. “We wanted similar though shorter engagements, and EBA came to the fore.”

 

In each 3-day EBA, Parameta’s team, supported by AWS experts, develops new solutions to business problems. A preparatory phase of 3–4 weeks precedes each EBA, during which Parameta defines goals and tasks and sets up the logistics for people to work on coding and development during the EBA. “We cut away all the red tape in the EBA, and as people from different functions work together, we find that they reinforce each other,” says O’Gorman.

During one of the EBAs, Parameta’s team used Amazon Bedrock—a fully managed service that offers a choice of high-performing foundation models—and call and chat records of interactions among financial trading desks to build a generative AI solution. The team built the data pipelines, Amazon Bedrock orchestration, and evaluation framework and added in observability to create a production-ready solution. The solution can extract order, pricing, and trade data from chat records and build insights. Monetizing this dataset can create new revenue and customer growth opportunities for Parameta.

 

Parameta used the experience gained in the EBA to develop other business solutions that fulfill customer needs. For instance, the data science team used Amazon Transcribe—a fully managed, automatic speech recognition (ASR) service—to automatically extract text and layout elements that built an energy and commodities dataset that was not previously available. Parameta sold this dataset to a buy-side client, who began trading with TP ICAP, resulting in new data and trading revenue. Similarly, the company’s operational efficiency increased as it automated the processing of end-of-day pricing for commodities. Parameta also used Amazon Textract, a service for automatically extracting printed text, handwriting, layout elements, and data from any document.

 

After using serverless technologies during the EBAs, Parameta increased serverless adoption to gain scalability and cost advantages. “We can focus on the business and let the technology handle the infrastructure,” says O’Gorman. “If the volume of incoming data increases, the serverless technology stack scales automatically to handle it.” Using serverless technologies, Parameta pays only for the resources it uses, which makes it simpler to allocate costs against revenue. Serverless technologies also help Parameta reuse components across solutions and use cases, shortening the development cycle and speeding up time to market.

 

Parameta uses EBAs to create solutions that enhance customer value, but their most far-reaching impact is on the company’s internal teams. “Relationships between teams improve as silos break down in EBAs,” says O’Gorman. “People work together with intensity, commitment, and focus, and they find the EBA an amazing experience.” The teams upskilled themselves too, learning the “Working Backwards” approach as well as upskilling on technologies like generative AI.

Parameta will continue to use EBAs for faster solution development, looking to capture opportunities that directly generate revenue and solutions that surface monetizable data. For instance, it plans to extend the solution that extracts data from chat records to include chat transcripts from other sectors or markets.

 

The company also wants to run EBA-like events on its own. “One of the biggest wins from doing an EBA is the learning that we’ve taken forward,” says O’Gorman. “AWS has given us the tools that we need to work in an EBA-like event by ourselves.”

 

The success of EBAs and the excitement of the people involved has led Parameta to consider making it a more widespread way of working. “There are some cultural aspects we can change so that every day can be like an EBA—as much fun and as productive,” says Roland Anderson, chief information officer at Parameta.