Vertex are a publicly quoted (NASDAQ) US-based company that provides end-to-end tax software solutions for its customers, enabling them to manage compliance accurately and effectively. Vertex provides its range of solutions. to customers globally. The company has been in business for almost 50 years.

Tax law is complex, with expert vocabulary and inter-dependencies. It is also regularly updated. Timely access to changes in tax law across different jurisdictions and industries is a key challenge for organisations, and one that Vertex aims to tackle. Their portfolio already included several products in that area, and they wanted to create a new one focused on automated summaries of tax laws related to Sustainability.

Vertex engaged with Spark to expand their data and AI solution around Sustainable tax laws. Spark initiated an R&D phase with Vertex to prove the feasibility of an automated tax law summaries using: web scraping and filtering of applicable laws, key points extraction and summarisation. The initial focus for the purposes of proving the concept was on the UK market, which is highly complex.

In the competitive domain of tax assistance services, Vertex initiated a new project in order to make tax law more accessible to their users. They initially started to develop an early version of the platform in-house. After a number of months of work, they engaged Spark’s AI and Machine Learning team to deliver a law summarisation prototype, as part of their research and development effort. The Spark-developed component was required to deliver both sourcing of relevant content, and subsequently the summarisation of this content into a format appropriate to Vertex’s end user customers.

The primary outcomes delivered by the Spark team during this project were:

Fast extractive summary

• 5 minutes processing between law ingestion from original website to summary.
• Empowered Vertex redactors to quickly review and redact summary for public use.
• Highlighted important parts of the law.
• Provided Vertex redactors with contextual information about the law.

A scalable solution. Spark designed a modular architecture allowing Vertex to seamlessly add more law sources over time:

• 3- 4 weeks to create a new connector.
• Horizontally scalable.
• Ability to process hundreds of laws in matter of minutes.
  
  

Ability to extend across other business lines:

• The system was designed to work with other types of laws and languages.