How to decide whether to build or buy fintech, AI technology

The decision on whether to build your own technology platform or program versus through a partnership is becoming more complex, and oftentimes more costly, as technology has advanced into machine learning and artificial intelligence.

Any learning model requires more data and updates to perform better, which also raises security concerns about data flows and whether to develop technology in-house or work with a tech partner to streamline that process. 

READ MORE: The ROI on AI: Advisors struggle to get unbiased answers from tech providers

“That’s the multimillion-dollar question: To build versus buy,” said John Mackowiak, chief revenue officer at Advyzon, a cloud-based portfolio management platform based in Chicago.

Mackowiak said his company builds technologies in-house mainly to create a consistent user experience, but also because adding vendors would require integrations into several different systems, which can be tricky. 

“Certainly it [building in-house] was not the easiest path to go down. But over the decade since we launched, it has changed the game,” he said. “Because with the idea of integrating vendor to vendor, you end up with a fragmented tech stack and a fragmented user experience.”

But building new technologies in-house is not always right for everyone. Every firm has a different business strategy, budget, client base and varying level of capacity for technological development.

“You have to really understand your use case really well. What are the economics behind it? And then, AI, just like any other technology, its capital,” said Lee Davidson, chief data and analytics officer at Morningstar. “Usually, you’ve got capital and labor to deploy to solve a problem. And you’ve got to find the right mix to get the production function to work, meet the client’s requirements.”

Davidson gave an example of learning from “failures” when they began experimenting with machine learning models in 2011 to generate investor insights for users. 

“They never went anywhere” because “I thought as a researcher working on that, that people cared about accuracy,” he said. “They cared about transparency.”

Morningstar pivoted to using machine learning and AI-based large language models to better explain to users how an answer was formulated.  

“What we found was there’s this trust factor, this explainability, so we started wrapping around it more explainable pieces” he said. So the user tool “talked about the fund, talked about the context … and we got a ton more usage because people were understanding there’s some evidence behind it, some context to it.”

READ MORE: Is your chatbot speaking the right language to your customers?

Morningstar has also been deploying new tools through large language models, like ChatGPT, to help launch its AI research assistant Mo last year. Increasingly often, a firm might build one system internally but outsource another feature or tool to a more specialized tech company. 

“It’s hyper-individual. For us, we have our in-house team that designs what we’re doing from an AI perspective. But we do use third-party tools to run the system,” said Chris Shuba, CEO of Helios, a quantitative asset management platform that uses learning tech to perform portfolio data analytics on thousands of mutual funds and ETFs. “Some people don’t do that at all. They just say to an AI developer, ‘Here are the results that I’m looking for. Go do it and we’ll pay you for the project.’ Other folks could have some hybrid.” 

READ MORE: How the relationship between financial advisors and technology is changing everything

For example, Envestnet is a large wealthtech provider that offers technologies built in-house, but it has also partnered with specialized tech providers like iCapital to offer an Alternatives Exchange on its platform for client advisors. 

Dana D’Auria, group president of solutions and co-chief information officer at Envestnet, said the company chose to work with iCapital and others who were “established providers” specific to the alternatives space. 

“That’s just not something that we’re going to build in-house because No. 1, it would be an enormous, distracting task,” she said. “And No. 2, you already have leaders in the space who are in the same ecosystem of clients that we serve. So that’s an obvious one to partner with.”

READ MORE: Envestnet, Salesforce join forces on advisor platform

D’Auria said when considering whether or not to build in-house, firms should first gauge what the build will look like, whether pulling resources internally would be distracting from overall objectives or if internal staff has better knowledge of the business and integration. 

“Bringing a build in-house takes resources, but at the same time … you’re getting the same team who works on the proposals, the same team who works in the billing environment — all those same people are now embedding it directly into the solution.”