How Belinda Neal helps keep Goldman Sachs’ engineering team on track with scalable guardrails

As chief operating officer for core engineering, head of product management and head of engineering partnerships at Goldman Sachs, Belinda Neal helps define the core engineering strategy at the firm. She’s also responsible for technology business development, bringing innovative tech solutions directly to the lines of business and part of the team that defines Goldman Sachs’ AI strategy.
It’s a ton of responsibility at a huge financial institution with over 45,000 employees, a market cap of almost $200 billion and close to $3 trillion in assets under management. While she is eager to introduce new technology to the firm, she is also part of a highly regulated company, and as such she has to move more carefully than a counterpart at a company dealing with less oversight.
That means she must look at new technology from both a functionality and safety perspective, carefully considering both, while always laying down a firm security and governance foundation, regardless of the technology she is trying to implement.
“We take governance extremely seriously. Security is our top priority in general and as we employ and develop use cases with AI in particular, they have to go through this rigorous review,” Neal told FastForward. She says the key is to start with small, controlled groups of users, see what works and then scale it to a broader group as they see success and the solution evolves.
That deliberate approach enables Goldman Sachs to put new solutions in place in a controlled way, while maintaining a close working relationship with the business units. It’s similar to how Databricks CIO Naveen Zutshi sees it. “I think where a lot of IT departments fail is that failure to have that tight communication with the business operations,” he told FastForward earlier this year.
A careful approach to AI
Neal says for Goldman, that means developing tech strategy through a product and value lens. “We think holistically as our technology strategy evolves in the new age of generative AI that we find ourselves in. Ultimately, we want to really drive value to the business through the tools and through the technologies we're developing,” she said.
She sees AI as a way to make the business run more efficiently. “AI allows us opportunities to create more efficient business processes. It allows us to enhance our client experience, and then more broadly with the developers, it allows us to make our engineering groups and communities much more effective at developing technology at a rapid rate,” Neal said.
A prime example of that is the GS AI assistant announced recently and rolled out to 10,000 employees for starters. “The GS AI Assistant was built on top of our GS AI platform, and the goal was to enable our employees to access large language models in the right way across the firm,” she said. That involves putting guardrails in place to keep questions and answers confined to areas that make sense to a bank. The responses also include links to the source of the answer to let employees confirm the accuracy and develop trust that the assistant is providing appropriate and authoritative responses.
They built the assistant on top of an internally-developed AI platform with the security and governance they require, while also providing the flexibility to plug in different models as the market changes. “We wanted to build this platform in a way that allowed us to innovate fast with the best models on the market. So, creating this AI platform and then this AI assistant will afford us to be able to race very fast with the latest and greatest models that develop on the market while ensuring safety and security,” she said.
Looking to the agentic AI future
As with many large organizations, Goldman Sachs is taking a long look at agentic AI, knowing that this is where this is all heading. “The future state involves the agentic capabilities of the AI platform. That will then be able to allow people to create more efficiencies in how they operate,” she said. And eventually that will culminate in tools that do things on behalf of the employee or client.
As companies begin to have agents undertaking tasks independently, especially inside a bank, that is going to require governance, guardrails and structure to do it safely. “We're going to have to start thinking about how we manage and monitor agents in the same way that we would other components of our technology strategy,” she said. In some ways, it will be even more important as agents behave like employees and undertake tasks on behalf of customers.

Part of staying on top of the shifting tech landscape is being aware of what’s happening with startups. She says a big part of that is establishing relationships with venture firms to understand what’s coming and then building relationships with innovative companies.
“We are working very closely with new and interesting companies building really exciting new technologies. And we love engaging and learning about new and exciting startups. I'd say that we always have our ear to the ground to understand where the opportunities are,” she said. That includes working closely with the venture community and keeping an eye on the startup community to see how the bank can take advantage of new and innovative solutions.
When it comes to innovating in-house, she’ll continue to work within the organization to help the business take advantage of AI and other technologies as they develop in the coming years. “I think helping people adopt this new technology is going to be an important part of its success, and we’re going to work to help people find ways to leverage LLMs and other technologies in ways that will really add value,” she said.
Photo courtesy of Goldman Sachs