Banks have spent years managing technology implementations through procurement processes, vendor reviews, and controlled rollouts. The rise of forward deployed employees (FDEs) is introducing a very different model: AI providers are placing their own experts inside financial institutions to help build and deploy systems alongside the teams that will use them.
These embedded employees represent a deeper level of involvement than a typical technology partnership. Frontier AI providers are not simply delivering a tool and stepping away—they’re taking a hands-on role in how agentic AI is designed, integrated, and governed. OpenAI has even developed its own deployment company dedicated to sending these employees and engineers into the companies adopting its technology.
For financial institutions, this approach creates new questions around accountability, governance, and workforce composition. A new report, Agentic AI and the Rise of Forward Deployed Employees in Financial Services, by Matthew Gaughan, Tech & Infrastructure Analyst at Javelin Strategy & Research, looks at the challenges of incorporating these experts into an industry where technology adoption must operate within a highly regulated and risk-conscious environment.
Skin in the Game
Forward deployed employees are not necessarily a new concept. Software companies like Palantir have utilized forward deployed engineers in the past, although recent announcements and job postings have included roles that extend beyond engineering.
In the latest partnerships, such as those between Anthropic and FIS or between OpenAI and Fiserv, deployed employees may have more skin in the game than previous models have required. Executives at the vendors where they are deployed have a significant stake in seeing these efforts succeed.
FDEs are embedded within organizations for a specific purpose. With FIS and Anthropic, the initial focus of their relationship will be financial crime, with the hope of eventually expanding into other areas. Many of the companies where AI experts are being embedded don’t have the resources or employees capable of handling these initiatives internally. FDEs generally have more authority than traditional consultants, who often focus on evaluating potential projects for the benefit of management or the broader organization.
“In that sense, these FDEs may have a little more agency,” said Gaughan. “The executive suite has a vested interest in these things going well, but also there’s a lot more money behind it.”
Building Relationships
Financial crime makes sense as an area where these relationships can begin because it can help financial institutions make better decisions about whom they lend money to and how much money they lend, while supporting value-added solutions through improved information. The announcement between FIS and Anthropic said that Anthropic’s frontier model would serve as the reasoning engine within the financial crimes solution.
“That’s the type of relationship you’ll see,” said Gaughan. “These frontier AI labs are coming in and not necessarily building new solutions or new products out of whole cloth, but strengthening existing solutions and making them more efficient, more capable, faster, and able to handle more. For example, a model from one of the leading frontier AI labs may have capability beyond what the bank currently has to source different types of data, or to find different connections between certain types of data that may not have been surfaced previously.”
A bank looking to implement its own FDE model needs to understand how it could change the organization, especially if it adopts a framework that gives FDEs more agency over what is designed and built. An existing employee responsible for a particular area may find their role shifting as an FDE gains access to executive leadership and significant influence over development decisions. Questions around organizational structure will likely emerge as this approach becomes more common.
It’s important to view this as an evolution. A key part of that understanding where the company stands technologically and organizationally and ensuring that the right structures are in place to accommodate cutting-edge technologies like AI.
Clash of Cultures
There can also be a culture clash resulting from these embedded employees. Technology companies often comes from a “move fast and break things” ethos, which can conflict with banks that have operated under a stringent regulatory framework for nearly two decades following the financial crisis. Overarching all of this are broader questions about the implications of AI and how it will affect work, which remain hot-button issues.
“Beyond just the implications of adding forward deployed engineers and how that might change decision structures and reporting structures and responsibilities, it’s an area that both sides will need some education on,” said Gaughan. “The employees coming in from these companies will have vast expertise in AI, the models that they’re using and how they can be applied within certain use cases. But on the flip side, banks have all this experience working within a strict regulatory environment, including the compliance requirements and the internal things that go along with that.”
As a result, bringing on FDEs will likely require new training and development initiatives on both sides. Bank employees will need to understand how these agentic AI solutions will impact the business lines they support or the processes they oversee.
“The employees within these banks have experience that is relevant to making sure that these things are being applied in the right way that fits within that regulatory framework,” said Gaughan. “The organizational change is not just moving around people on a chart. It’s a little deeper than that.”
Here to Stay
Events are moving swiftly in this area. Since the publication of the report, many companies have changed how they are deploying AI. Frontier AI Labs have also adjusted their pricing models, and companies that have been using these solutions internally are facing higher-than-expected costs.
“It’s still pretty early innings on this right now,” said Gaughan. “I do think the forward deployed engineer framework is here to stay, at least in the near term. It’s how these frontier labs see themselves distributing their solutions more readily and stickily within some of these companies.”
“There’s questions that come along with integrating these types of technologies into your organization,” he said. “A lot of these are concerns that come with any technical decision. The important takeaway is knowing what your customers want and need.”
