Amid the hype surrounding agentic AI and its potential to transform financial services, firms across the sector have raced to adopt the technology. However, additiv founder and CEO Michael Stemmle believes the real differentiator is not how much AI you can implement, but how it is used.
Over a remarkably short period, AI capabilities have grown exponentially while costs have fallen dramatically, making AI tools accessible to firms of all sizes. However, having multiple AI tools is not the best way to generate the most value from them.
When operating in a regulated environment, firms looking to automate workflows through AI need to have deterministic rules that can be applied to ensure they handle tasks correctly. These models will need to have real-time access to data, and predefined frameworks around what the products are, business logic, pricing, risks, regulatory requirements and more. This information must then be made accessible across business functions and throughout the entire client lifecycle. To get the most value out of AI, firms need to wire this deterministic financial logic into the AI agents and orchestrate them to automate workflows in an expected way, providing reliable support to their human counterparts. Stemmle said, “You have to orchestrate that, otherwise it doesn’t work.”
What makes agentic AI such an exciting prospect for financial services is its ability to take automation to greater levels. While the digitalisation of the industry has helped to improve workflows by reducing manual workloads, such as automating data entry or collating information across datasets, agentic AI is able to go further by handling scenarios that need human-like logic. Instead of being a simple tool to handle a specific task, agentic AI can use reasoning across processes, enabling them to manage regulated processes end-to-end. Stemmle noted, “You can now offer end clients a completed experience and replace manual human processing work.” However, it is important to note, the technology is not to replace the human, as while the technology is powerful, it is unreliable, especially for regulated environments.
Instead, the argument to be made is that the winners in the coming years will not be the firms with the most sophisticated models, but those that can safely route work between AI agents and people, ensuring transparency on each decision and run it on their existing core systems. He said, “AI itself is not a strategy; it’s a tool. However, it allows for a new operating model or strategy, which haven’t been possible before or feasible economically.” However, when the technology, platform, and orchestration layer are combined, firms can deepen their services and “gives you a completely new starting point in the competition and in creating the services for your clients,” he said.
additiv’s commitment to orchestration
Orchestration has been central to additiv’s strategy from the outset. When the new digital capabilities were becoming common across financial services, they understood there would be new challenges and orchestration was the way to overcome them. While many firms have focused on protecting legacy core systems or adding AI bolt-ons, additiv saw a better route forward was by serving the layer in between. This was a specific orchestration layer that could connect AI to existing rules-based financial logic, without needing to replace anything.
Stemmle explained, “I wouldn’t underestimate that orchestration is nothing else than platform strategies, from a technology and business point of view. If you look at what Revolut is doing, because they had a pre-built approach. They have a platform where they can define services offering businesses, then distribute it, and then decide what they do by themselves and what they source from third parties. We give them full flexibility and control over the value chain and greater speed to the market and being much more client centric.
“That’s a typical orchestration platform. That’s the core enabler for them to do business. Financial service providers have learned if they would like to adhere to the same strategies or same operational advantages as, for example Revolut has, they can use our platform.”
One of the most common scenarios when it comes to trying to adopt generative AI tools is to just bolt on solutions to their existing infrastructure without that orchestration layer. This only leads to a dead end, where the technology cannot function correctly. For the technology to be acceptable within a regulated environment, such as financial services, it needs to be explainable, auditable and accountable, which can only be achieved through orchestration, deterministic logic and human-in-the-loop. When it is done correctly, the AI will streamline processes, ensuring better decision making, lower costs and quicker turnaround times.
To illustrate the point, Stemmle said, “I’ll make a comparison. When the PC came, if you would have taken away the typewriter and put the PC or the computer on the table, but kept everything the same, the productivity gain would be zero. You have to rethink how you distribute, how you operate, how you service, how you source, and so on.”
What the future looks like
Agentic AI will reshape how financial services operate. While many fear they could spell the end of the human worker, that is not the case. Instead, they will act more like a partner. These agents will absorb the repetitive processing, while the humans will be free to tackle more creative tasks, Stemmle noted. However, people are reluctant to change, and this could see firms sitting on the sidelines and waiting for others to pave the way. That is not the right path, firms need hands-on experience.
“For the first movers, it’s not just about having this stuff in place, it’s having this learning curve. This is like an iterative process where you start and learn, and the learning curve and experience is a huge advantage, and the sooner you start, the better.”
This does not mean firms should start from scratch and redesign their whole company but start small with something that works and keep expanding and experimenting. For instance, an idea that appears promising today may prove ineffective after several months of testing. That is needed to ensure firms can keep pace with the rate of innovation with AI. “The ones who have a playful approach and learning attitudes are going to succeed,” he said.
Agentic AI has been widely available in the market for under two years, and it will continue to change rapidly and gain more impressive capabilities. This is not going to be an optional development, in fact, Stemmle describes it as a generational change that will have seismic consequences, much like the advent of the steam engine. As such, those running a blended workforce with humans and agents, with an effective layer of orchestration will be best positioned to compete. Those that implement disjointed AI tools could be at risk.
On a final note, Stemmle concluded, “I would say that in three years, a financial service provider that is not working on a platform where they can define, govern, and execute agentic workflows will not be competitive anymore. You can still use a horse to go from Southeast London to Piccadilly Circus, feel free to do so, but it’s outdated.”
additiv was recently named in this year’s AIFinTech100, which identifies the companies leading AI solution providers companies helping to transform financial services. The full AIFinTech100, including profiles on each company, can be found here.
Copyright © 2026 FinTech Global
