Money manager VanEck is backing an under-the-radar AI fintech to boost its ETF business

FinChat
- New York-based VanEck is partnering with a generative AI fintech and backing it with a $1.5 million investment.
- FinChat is a Toronto-based startup that offers generative AI software for investment research.
- A VanEck exec outlines how the money manager will use FinChat and why it invested in the startup.
The $119 billion money manager VanEck just threw its weight β and capital β behind a little-known fintech startup asΒ the generative AI race heats up on Wall Street.
The New York-based investment firm is onboarding AI-powered investment software from Finchat for its employees to use and making a $1.5 million strategic investment into the startup through its venture business, Wyatt Lonergan, VanEck Ventures' general partner, told Business Insider.
FinChat is a Toronto-based global equities research platform. Much like ChatGPT's chat interface, investors can type questions and tasks into FinChat. It uses public and private data sets to generate answers that are used to create documents, charts, and presentations, among other Wall Street work staples.
Lonergan is betting FinChat will save his VanEck coworkers time and give them a leg up selling products like equities ETFs. He added that the company is also considering leveraging the tech in its crypto funds.
"What would previously take an analyst 30 hours to do, now you could write a prompt, have it go pull that data, and create a presentation for you" in about 30 minutes, said Lonergan, who joined VanEck from crypto fintech Circle in July.

VanEck
Some use cases include building slide decks comparing VanEck's $22.6 billion semiconductor ETF with competitors' similar products. While on calls, salespeople can quickly answer questions in real time, like what the underlying assets in a specific ETF are.
Putting slides together used to involve pulling data, loading it into Excel, building a chart, and comparing that chart against another dataset. Then, employees would slot it into PowerPoint and send it to the client. FinChat also makes sharing this data and analysis with clients easier by aggregating that information into a link.
"This is where I think we're going. You're being able to pull these things in real time with very, very simple prompts, and I think that's the magic of AI," Lonergan said.
A closer look at how FinChat works
When FinChat cofounder Braden Dennis walked into the boardroom to pitch Jan VanEck, the firm's CEO, and other employees on his startup's generative AI software, he was pleasantly surprised he didn't have to do much.
"A few of the analysts, they took over the demo. They were showing the executives at the firm how they were really using it and loving it," Dennis told BI. For the next 20 minutes, the analysts showed the room exactly how he used the tech for the ETF business, Dennis said.
The startup, which launched in 2023, has made more than $2.5 million of annual recurring revenue and is growing about 15% month over month, Dennis said. He expects to close 2025 with about $10 million of ARR. FinChat has raised $3 million to date.
Behind the scenes, FinChat relies on several generative AI model providers, including Anthropic, OpenAI, and open-sourced models, to not be tied to just one company's progress. FinChat runs an evaluation test every week that compares the quality, speed, and cost of each model's response to common query types. Based on the test results, FinChat dynamically routes prompts to models with the best test results for that specific query.
For example, Dennis said FinChat found that Anthropic's model, Claude, is best at transcript summarization, whereas OpenAI is great at analyzing certain companies.
FinChat uses public data, like earnings presentations and SEC filings, and its own proprietary data, such as company-specific KPIs that break out specific revenue streams like ads versus subscriptions. Dennis said the startup is building a way for its platform to pull in enterprise clients' internal data, adding that the feature should be available by this summer. FinChat also uses data from third-party vendors.
The fintech also offers user-generated automated workflows for things like investment committee decks, one-page memos, or consolidating every income statement graph into one PDF.
Let's say it takes 15 prompts into FinChat for an analyst to put together an 8-page summary on why Spotify has a competitive moat. The analyst can save those prompts for the next time they need to do a similar report for Apple Music or Tidal.
The goal with FinChat is not to automate people's jobs away, Dennis said, but the redundant parts of their jobs, like data aggregation, the first steps of building models, and paperwork.
"These are not things that an analyst would say is what makes them great. What makes them great is the thinking, the decision-making, the client relationships," Dennis said.
"Let's save the paperwork for the robots and the thinking and collaborating for the humans," he said.