Paysafe Says AI Wins Depend on Culture, Not Code Alone

Enterprise leaders have spent the past two years in a race to adopt artificial intelligence (AI), often treating it as a procurement decision rather than a systemic transformation. Budgets have flowed toward large language models, copilots and automation platforms, with the implicit assumption that better tools will naturally translate into better outcomes.

Yet across sectors, the results remain uneven. Pilots scale slowly, productivity gains plateau and promised returns on investment can prove elusive. At Paysafe, however, the approach has been different.

“One of the biggest drivers for us was that our CEO was bought in very, very early. Almost four years back, he started talking about why this is important and how this is going to change everything,” Ahu Chhapgar, chief technology officer at Paysafe, told PYMNTS.

Rather than treating AI as a series of experiments, the payments company has embedded it across the organization, offering a revealing case study in what it takes to operationalize AI in the messy realities of compliance, culture and code.

The core issue surrounding enterprise AI’s scalability and adoption is not technological capability. Today’s AI systems are already powerful enough to reshape workflows, compress cycle times and augment decision-making. Instead, the constraint lies in how organizations are structured, how leaders align incentives and how work itself is defined.

Building AI in a Regulated Environment

Paysafe’s own early executive alignment did more than signal priority; it reshaped internal expectations. AI was no longer optional experimentation, but a company-wide mandate tied to outcomes.

That top-down clarity enabled what Chhapgar described as a coordinated transformation built on three pillars: culture, talent and execution discipline.

“Culture drives the will across the company,” he said, emphasizing that enthusiasm alone is insufficient without the right expertise and partnerships, especially in a sector like payments, where regulatory scrutiny is constant.

“We are highly regulated. Risk is super important. Compliance is super important,” Chhapgar said, noting that the challenge is not simply introducing AI, but doing so without eroding the safeguards that underpin trust in financial systems.

Introducing AI into sensitive processes can raise new questions: How much autonomy should systems have? How does risk propagate? And where should human oversight remain?

“You have to make sure that you build systems and processes that do not create a risk of skipping all the important checks that you actually needed,” Chhapgar said, stressing the importance of testing frameworks against existing processes before scaling them, and treating governance as an evolving system rather than a static checklist.

The New Developer Workflow and Expanding Business Case for AI

AI at scale is a management and operating model transformation, not a technology rollout, and its real payoff is new business creation, not just efficiency. The impact of this shift is increasingly becoming visible across engineering.

“Things that took a long time before are now being done at a very fast pace,” Chhapgar said, describing an organization where the “excitement … is absolutely palpable.”

Yet the gains are not purely about speed. The nature of engineering work itself is changing. Developers are becoming orchestrators and defining problems, constraints and system design while AI handles much of the execution. This shift, however, can introduce new bottlenecks. Code reviews, for example, have become more intensive because AI-generated output tends to be verbose.

“It is more about setting context and setting architecture,” Chhapgar said.

Beyond productivity gains, AI is reshaping what the company can build. Historically, many product ideas never made it past the business-case stage due to cost constraints. That calculus is changing.

The implication is significant, as AI is not just optimizing existing workflows but expanding the frontier of viable products and markets. This alignment ensures that AI is not confined to the technology organization but becomes a shared responsibility across the business.

“We have goals for every single functional group to show AI-related gains,” Chhapgar said. “While experimentation is really, really important, what we care about is real value addition.”

That value typically falls into two broad categories: optimization, such as cost reduction and improved fraud detection, and expansion into entirely new product domains enabled by lower development costs. This expansion can be viewed as a particularly consequential one in the context of agentic AI, or systems that can act autonomously on behalf of users.

The Emerging Enterprise Playbook for Agentic AI

In payments, where trust is paramount, the stakes for agentic AI are high.

Chhapgar framed the challenge succinctly: “The most important thing is to establish trust between all parties.”

As AI agents begin to initiate transactions, questions of liability, accountability and fraud prevention become more complex. Who is responsible if an autonomous system makes a purchase? How are disputes resolved? These are not abstract concerns but foundational issues for the next generation of digital commerce.

Paysafe’s own approach is to embed simplicity at the user level while managing complexity behind the scenes. Payments have always required this duality of intricate infrastructure paired with frictionless experiences. AI raises the bar but does not change the principle. The company is working with major technology firms and card networks to define how trust operates in an agent-driven ecosystem, signaling that industry-wide collaboration will be essential.

In the end, AI is not a shortcut to transformation but a catalyst that exposes the strengths and weaknesses of an organization’s underlying systems. The enterprises that recognize this and react accordingly could be those that define the next era of competitive advantage.