Data models: an engine for growth
9 November 2020: Engine B company co-founder Shamus Rae explains how the open source project aims to bring consistency into client data in fields as complex and diverse as audit, tax and the legal sector.
Engine B has its origins in a series of research projects at KPMG. Its co-founders, Shamus Rae and Donne Burrows, were working with Imperial College to look at the impact of artificial intelligence on the audit, tax and legal sectors. “We actually broke down audit into 18,000 different steps, so it was a detailed piece of work,” Rae explains. “In doing so, we realised that a significant amount of audit could be automated.”
Off the back of their research, Rae and Burrows’ project team built a five-year automation plan encompassing several KPMG services. As part of that, Rae and Burrows started looking at the blockers that halted transformation. One of the biggest was around client data.
“Client data access was costing a lot of money for the Big Four but was also creating a moat in terms of competition,” says Rae. “It meant that technology companies trying to bring innovation into audit and tax had to create feeds into hundreds of different data sources. Which gave us an idea.”
That idea was the seed of what would become Engine B: an open-source data company that is working with a number of organisations – including ICAEW – to create a set of data models and dynamic knowledge graphs that will make gathering and sorting data much quicker and easier for accounting firms and organisations. It started with a conversation with several audit firms to define common data needs, which in turn would help create a data standard for audit and tax.
By August 2019, it became clear that the firms could work together on an audit Common Data Model. After consultation with regulators, Engine B was born: a cross-industry project working with an initial nine audit firms (since increased to thirteen). ICAEW is a shareholder in Engine B with a place on its board to ensure that members get a voice in the development of the organisation’s data models. Other non-audit partners include Microsoft, Imperial College and the University of Birmingham.
“It's a cross-industry project,” Rae explains. “It is a for-profit company, but it is purpose-driven. ICAEW is there to protect the industry and make sure that our purpose – opening up the professional services market, creating these open source and common data models and enabling automation – is not competing with the industry.”
The knowledge graphs element of Engine B’s platform, when combined with the audit Common Data Model is a potentially game-changing duo. At a high level, knowledge graphs are all about context and relationships. They use interlinked sets of facts that describe entities, facts or things and their interrelations in a human and understandable form. Using knowledge graphs with the audit Common Data Model reduces the process of cleaning up and organising data from days to about an hour. “It means that for data analytics teams in the firms, there's a whole chunk of work that they don't have to do anymore,” says Rae.
This work has been driven by industry needs and is a truly collaborative approach – a unique position to take within the sector, according to Rae. “People have tried to do data models before, but they now see that this is more collaborative with more firms and building the assets to make it work for the knowledge graphs and data feeds. Also, to be honest, having Microsoft behind it as well is not a bad thing, in terms of that broader data models perspective.”
The make-up of shareholders may seem unusual, with so many audit firms and associations owning a stake in the company, but the investor structure is very deliberate, says Rae. For one thing, the shareholder firms provided IP to develop Engine B. It also helped to cement the truly collaborative, cross-industry nature of the project. On the other hand, with a limited number of firms on the Board, there was a risk that the wider sector wouldn’t be represented – which is where ICAEW’s stake in the company comes in. “We are about opening up the market. We want ICAEW to represent the long tail and keep us honest,” says Rae.
The next step for Engine B is to build up its knowledge base. The company is seeking more organisations to collaborate with, looking to use their IP to help build more sector-specific models.
“We really want to try to crowdsource intellectual property around the systems people need us to connect into,” says Rae. “For the Big Four, there is a minimum of 300 different systems they've got to feed into, but you can go into the tail, it's going to be thousands. We want to do more. We will need to leverage the knowledge of those systems from the crowd and build the templates and the feeds. We're really keen to work with people to do that.”
The company also wants to find more firms to test its data model, and it is also working on common data models for the legal, insurance and tax sectors. But ultimately, 2021 will be about rolling out its assets to firms and companies. “That will be in the UK first, then the US. Then we'll roll it out in the rest of the world,” says Rae. “We've got a lot of pull from Latin America at the moment, and Singapore, Indonesia, Australia, etc. We can't lose sight of the base here in the UK – the government's put money into this – but there is scope to take it quite far.”
To contribute to Engine B’s research, email hello@engineb.com