Tsvetomir Doskov, Senior VP of BFSI at Sirma Group, was featured in an interview on Bloomberg TV Bulgaria’s “Business Start”. He shared his vision for the future of payments and discussed the current value of AI in the payments sector. Doskov explained that while there is significant automation in payments, true autonomy is still further away than many in the market expect. Additionally, he highlighted how account-to-account payments and mobile wallets are gradually changing user behavior, even as credit and debit cards remain the dominant payment method, in terms of volume if not in functionality. Here are some key takeaways from the discussion.
The conversation surrounding artificial intelligence in financial services has become increasingly intense. Discussions about AI agents transforming areas such as customer service and risk management are everywhere. However, when considering payments as an essential and highly regulated function within banking, I find that the reality is far more complex than the hype might imply.
My vision is grounded in a clear perspective on where AI truly stands within the payment ecosystem. While automation is advancing rapidly, genuine autonomy remains a distant goal. Here’s what I believe BFSI leaders, FinTech innovators, and enterprise decision-makers need to know.
The Current Reality: AI as an Enabler, Not a Decision-Maker
Despite the buzz around autonomous AI agents, I am unequivocal about the current state of AI in payments, that we are still far from being able to say that real agents can manage our payments.
Today’s AI systems in financial institutions primarily serve as support tools, such as sophisticated assistants that help human experts process vast amounts of information more quickly and accurately. They operate largely behind the scenes, enhancing speed and efficiency in various processes, but they do not make independent payment decisions. This distinction matters to me. In an era where AI agent has become a catch-all term, I want to remind everyone that not all automation equals autonomy. I see financial institutions deploying AI to:
- Analyze transaction patterns at scale
- Detect anomalies and potential fraud in real-time
- Accelerate compliance checks and reporting
- Support customer service with intelligent chatbots
But the final say, whether to approve, reject, or flag a payment, remains firmly in human hands, and this is for good reason.
Why True Autonomy Remains Elusive
In my view, payments sit at the intersection of technology, regulation, and trust. Unlike recommending a product or optimizing a supply chain route, payment decisions carry direct financial liability, regulatory scrutiny, and reputational risk.
My comments reflect a broader industry consensus I share: while AI can optimize how a payment is executed, the decision whether to execute it requires human judgment, contextual understanding, and accountability that AI cannot yet replicate.
This cautious approach is not a sign of technological limitation alone, it’s also a strategic choice I advocate for. In highly regulated environments like banking, explainability and auditability are non-negotiable to me. AI systems that operate as “black boxes” simply cannot be trusted with autonomous payment authority.
The Evolution of Payment Infrastructure
While AI’s role in decision-making remains bounded, the underlying payment infrastructure itself is undergoing significant transformation. For me, there are two key shifts:
1. Card Schemes Are Becoming Interface Layers Traditional card networks (Visa, Mastercard, and their regional counterparts) are, in my view, gradually evolving from being the primary payment rails to serving as interface layers that connect users to a broader array of payment services. This shift enables more flexibility and choice in how transactions are routed and settled. 2. Alternative Payment Methods Are Rising Mobile wallets, real-time payments, and account-to-account (A2A) transfers are gaining traction, driven by consumer demand for speed, convenience, and lower costs. I believe these alternative methods are reshaping the competitive landscape, challenging the dominance of traditional card-based systems.
For BFSI leaders, this means payment strategy can no longer be about optimizing a single channel; it’s about orchestrating a multi-rail ecosystem where AI can help route transactions intelligently based on cost, speed, and user preferences.
The Near Term: Scenario-Based Automation
So where is AI in payments headed next? I believe the immediate future lies in scenario-based automation. In this model, users provide advance consent for predefined payment scenarios, such as recurring subscriptions, utility bills, or threshold-based replenishments, and AI systems handle the execution automatically within those boundaries.
Think of it as smart automation with guardrails:
- User sets the rules: “Pay my electricity bill automatically every month, but only if the amount is within 10% of last month.”
- AI handles the execution: Monitors bills, validates amounts, processes payment if conditions are met.
- Human intervenes on exceptions: If the bill spikes unexpectedly, the system flags it for review rather than proceeding autonomously.
This approach, in my opinion, balances efficiency with control, delivering tangible benefits without crossing into the controversial territory of fully autonomous financial decision-making.
The Long-Term Vision: Intelligent Optimization, Not Autonomy
Looking further ahead, I picture AI as an intelligent optimizer rather than an autonomous actor. In this future state, I expect AI systems to be capable of:
- Selecting the optimal payment method based on context (e.g., choosing between a card, wallet, or A2A transfer depending on fees, speed, and rewards)
- Negotiating better terms dynamically (e.g., routing payments through the most cost-effective rail at that moment)
- Personalizing payment experiences based on individual user behavior and preferences
Therefore, critically, the decision to pay remains with the user. AI will recommend, optimize, and execute but not decide.
This vision aligns with a broader trend in enterprise AI that I subscribe to: augmentation over automation. My goal is not to replace human judgment but to amplify it with intelligent tools that reduce friction, minimize errors, and unlock new levels of efficiency.
Strategic Implications for BFSI Leaders
My recommendations include several important implications for banks, FinTechs, and enterprise software providers operating in the financial services space:
1. Adopt a Phased AI Roadmap Rather than chasing the elusive goal of full autonomy, I recommend focusing on incremental automation that delivers immediate value while building trust. Start with low-risk, high-volume processes and expand gradually. 2. Prioritize Explainability and Control In regulated environments, I believe AI systems must be transparent and auditable. Invest in explainable AI (XAI) capabilities that allow compliance teams and auditors to understand how decisions are made, even if humans are ultimately responsible for them. 3. Design for Multi-Rail Orchestration As payment infrastructure diversifies, I see AI playing a crucial role in intelligent routing. I encourage building systems that can evaluate multiple payment options in real-time and select the best one based on dynamic criteria. 4. Keep Humans in the Loop For high-stakes decisions, I insist on maintaining human oversight mechanisms. This is not just a regulatory safeguard, it’s a trust-building measure that reassures customers and stakeholders. 5. Align with Regulatory Expectations The EU AI Act and similar frameworks classify certain AI applications in financial services as high-risk, requiring strict governance and human oversight. I believe a measured, human-centric approach to AI in payments positions organizations favorably from a compliance standpoint.
The Bottom Line
The narrative around AI in payments is often framed as a binary: either we’re on the brink of fully autonomous financial agents, or AI is just a buzzword with no real impact. My perspective suggests a more accurate middle ground.
AI is already transforming payments but not by replacing human decision-makers. Instead, it’s enhancing efficiency, enabling smarter automation, and laying the groundwork for a future where AI helps users make better financial decisions without taking those decisions away from them.
In the financial industry, the key opportunity is not in pursuing autonomy for its own sake, but in using AI strategically to address real problems, minimize friction, and gradually build trust through automated solutions.
Watch the full interview here
