Episode summary
Building AI Systems for Capital Markets
Exchanges (Goldman Sachs)
· · 29 min · Hosts: Allison Nathan, George Lee · Guests: Chris Churchman
Summary
Chris Churchman argues that institutional AI products have to start from the investment workflow, not from the model. On Marquee, Goldman Sachs is building tools that help clients navigate research, trading commentary, analytics, and pre-trade analysis by turning a user’s intent into a grounded, auditable answer. He says the shift in software is from users learning systems to systems learning users, but for finance that only works if every output can be traced to a verified source.
A major theme is the gap between a flashy demo and a durable product. Churchman says demos look good on their best day, while products are judged on their worst, so developers should not waste time designing around temporary model limitations. Instead, he recommends building for what the model cannot learn on its own, such as firm-specific entitlements, mandates, and data relationships. He is skeptical that grounding alone will eliminate hallucinations, and he sees the field moving through prompt, context, and agentic engineering toward environment and mandate engineering.
Churchman also warns against outsourcing reasoning to machines. He frames the goal as augmenting human judgment, not replacing it, especially in a business where people must stand behind recommendations under uncertainty. He describes two paths for innovation: automation, which speeds up existing processes, and re-conception, which asks what new questions AI makes possible. His long-term optimism rests on scaling laws, synthetic world models, and the belief that the technology can still expand materially even as the form factor remains unsettled.
Key takeaways
- Chris Churchman says Marquee AI is designed to ground every sentence in a source or auditable calculation, because institutional users need provenance and trust.
- He argues that a demo is judged on its best day, but a product is judged on its worst, so teams should build for reliability rather than model workarounds.
- Churchman says developers should focus on what models cannot learn natively, such as entitlements, mandates, and firm-specific data connections.
- He distinguishes automation from re-conception, saying the biggest gains come from redesigning workflows around what AI makes newly possible.
- Churchman remains bullish on AI progress, citing scaling laws, synthetic world models, and the possibility that AI will help automate AI research itself.
Notable quotes
“a demo is judged on its best day, whereas a product is judged on its worst”
“you really have to build the framework to force it to ground itself because it inherently can't do it itself”
“Don't bet against the model, basically. That's the key”
“I think that my North Star here is, I feel a great responsibility building products for the firm”
How the episode opens
Welcome to Goldman Sachs Exchanges. I'm Allison Nathan, and I'm here together with George Lee, who is the co-head of the Goldman Sachs Global Institute. Together, we're hosting a series of episodes exploring the rise of AI and everything it could mean for companies, investors, and economies.
George, great to see you again. Good to see you. So, today we're digging into another fascinating topic: what it takes to actually build products in the AI age, the ingredients, the challenges, and the approaches behind building solutions that first and foremost actually work, but also actually matter.
And I really just wanted to first get your thoughts, George, because as we've discussed many times, you're talking to companies, firms, clients about these topics every day. So how much real progress do you see being made here? Well, I'm very excited for this discussion because in our prior sessions, we've spent a bunch of time talking about the foundational elements of AI, the infrastructural dimensions, and the topic of building applications in this environment I think is very timely and useful, and we have a great guest for that.
But look, I think there are a lot of questions in the ecosystem about what is the perimeter of the foundational dimensions of this relative to the application part of this stack. And so again, our guest will be able to untangle that for us in a really productive way. Right.
So let's bring in our guest. It is Chris Churchman. Chris is the head of Marquee, the firm's digital platform for institutional and corporate clients, and he's also co-chair of the firm's Global Banking & Markets AI working group.
So you're on the front lines, Chris, on all of these matters in terms of bringing new AI products to the market. So welcome to the show.
From the episode’s captions: roughly the first two minutes, lightly cleaned. Keep watching on YouTube →
Topics
- building ai products
- capital markets ai
- grounding generative ai
- demo vs product reliability
- auditable ai systems
Companies and tickers mentioned
- GS
- Goldman Sachs
Watch the episode
Watch “Building AI Systems for Capital Markets” on YouTube
Summary, takeaways and quote selection by Octagon. Quotes are short excerpts; the full conversation is in the episode.
More from Exchanges (Goldman Sachs)
-
Playing for Scale: The Future of Sports Investing -
John Waldron on GDP Growth, AI, and the M&A Outlook -
Brookfield’s Connor Teskey on the Real Bottleneck in the AI Buildout -
Goldman Sachs Chairman and CEO David Solomon on AI: Implementation, Accountability, and Jobs -
Why Oil Prices Are Surging Despite Recovering Supply -
Where Chinese Exports Are Gaining the Most Market Share
Exchanges (Goldman Sachs), this episode and its artwork belong to Goldman Sachs. Octagon is not affiliated with the show; this page is an independent summary.
