--- title: "Agentic AI Isn't Just About What It Can Do. It's About What We Can Trust It to Do." description: "Malin Gaertig reflects on Singapore’s Agentic Finance Summit: agent authority, accountability, trusted infrastructure and the gaps between discovery, selection and buying." date: 2026-10-10 dateModified: 2026-10-10 author: "Malin Gaertig" canonical: https://marksuite.ai/insights/agentic-finance-summit-trust-accountability-commerce categories: ["Agentic commerce","Governance"] publisher: MARK (https://marksuite.ai) --- # Agentic AI Isn't Just About What It Can Do. It's About What We Can Trust It to Do. *2026-10-10* The Agentic Finance Summit in Singapore highlighted a shift from building capable agents to building the conditions in which they can be trusted to act. For commerce, that means connecting discovery, selection and execution with reliable data, clear authority and accountable infrastructure. At the Agentic Finance Summit in Singapore, one point came through clearly: the challenge with agentic AI is no longer just what the technology can do, but what we can safely allow it to do. The hard part is no longer necessarily building the agent. It's building the conditions in which it can be trusted to act. ## Why is agentic AI technology moving faster than the frameworks around it? Mark Willis of Standard Chartered highlighted that software development is becoming cheaper, but the work required to establish compliance, assurance and confidence with risk teams and regulators remains difficult. For regulated financial institutions, that means progress is likely to be incremental. Start with controlled use cases, automate processes that already operate on established payment rails, demonstrate that they work, and expand from there. That isn't a lack of ambition. It's the reality of introducing autonomous decision-making into systems where mistakes have financial and regulatory consequences. This isn't unique to banking of course. Any business considering agents that can interact with its systems, customers or commercial processes will need to think beyond the capability of the underlying model. It will need to consider the data the agent can access, the actions it is permitted to take, the controls around those actions and what happens when the outcome is wrong. ## Who is accountable when an AI agent gets something wrong? One of the most important questions raised throughout the panel discussion was about accountability e.g. when an agent gets something wrong, who is responsible? An agent might have access to sensitive information, financial assets or the ability to initiate transactions. Giving it access is not the same as defining the limits of its authority. The discussion explored practical ways to constrain that authority, from temporary access and transaction limits to approved destinations, permissioning and the ability to audit activity. There is also a fundamental tension. The more human approval is required, the less autonomy the agent delivers which lowers its impact. But removing those checks before the right guardrails are in place creates a different set of risks. The objective isn't autonomy at any cost. It's knowing which decisions can be delegated, under what conditions and where intervention is still needed. That requires more than a capable model. It requires an operating framework and governance around it. ## Why is trust becoming part of agentic AI infrastructure? The conversations around identity, verifiable data, security and auditability were a reminder that agentic systems don't operate in isolation. Agents need reliable information to make decisions, ways to establish identity and permissions and the systems they interact with need to recognise, authorise and monitor their actions. The summit also explored the potential convergence of AI agents with tokenised assets and programmable financial infrastructure. Unlike many traditional financial processes, on-chain systems can support transactions and markets operating around the clock, with activity that can be inspected and audited. That creates exciting possibilities for automation while at the same time also raises the stakes when an agent can move money or deploy capital at scale. The opportunity is real, but so is the need for infrastructure that makes the activity controllable, traceable and resilient. ## What does agentic finance mean for discovery, selection and buying? For brands, retailers and the businesses supporting them, the same shift is beginning to take shape across the customer journey. First, AI influences discovery: which brands are surfaced, how they are represented and which options are recommended. Then comes selection: which brand best fits the customer's intent, based on the information available to the agent. And eventually, execution: can the agent complete the task, whether that means checking availability, applying the right terms, placing an order or completing a payment? These are connected stages, but they are not the same problem. A brand might be recommended but lack the accurate, accessible information an agent needs to evaluate it. It might be selected but have a customer journey that makes automated completion difficult. And even if the technical path exists, customers and businesses still need confidence in how the agent is acting. Being found is not the same as being chosen. Being chosen is not the same as being bought. That is why I see agentic commerce as more than an AI visibility challenge. It is a question of how discovery, data, digital experience, transaction infrastructure and trust work together. ## Which gaps should businesses close before delegating more responsibility to agents? The summit didn't suggest that fully autonomous finance will arrive everywhere overnight. The discussion was much more grounded: controlled deployments, practical use cases, evolving standards and the need to build a track record. But the direction is significant. As agents become more capable, the constraints will increasingly sit around them: access to reliable data, the systems they can interact with, the authority they have been given and the safeguards governing what happens next. For businesses, waiting until autonomous agents are commonplace to examine those dependencies risks leaving the work too late. The more useful questions to ask now are: Where could an agent already influence or execute part of our customer journey? What information and infrastructure would it need to do that successfully? Where would it fail, lose confidence or require human intervention? And who is accountable for the outcome? AI may influence what a customer chooses before the customer ever reaches your website. Gradually, it may also act on that choice. The businesses best prepared for that shift won't necessarily be those that deploy the most agents first. They'll be the ones that understand where the value lies, where the gaps remain and what needs to be in place before handing over more responsibility. ## Key takeaways - Agentic AI requires more than a capable model: businesses need the conditions in which an agent can be trusted to act. - Access does not define authority; transaction limits, temporary permissions, approved destinations and auditability help constrain delegated actions. - Controlled use cases and established payment rails offer a practical starting point for regulated institutions. - Being found is not the same as being chosen, and being chosen is not the same as being bought: discovery, selection and execution are distinct challenges. - Readiness depends on reliable data, usable infrastructure, clear safeguards and ownership of the outcome—not simply deploying agents first. Source: https://marksuite.ai/insights/agentic-finance-summit-trust-accountability-commerce ## See also (related MARK analysis) - [Just in: Why the Agentic Commerce Governance Gap Is the Real Risk for Brands Right Now](https://marksuite.ai/insights/agentic-commerce-governance-gap-real-risk-brands) - [What Australia's UCP Checkout Rollout Means for Brands Selling Through AI Agents](https://marksuite.ai/insights/australia-ucp-checkout-rollout-brands-selling-through-ai-agents) - [Just in: What Amazon's ChatGPT Ad Deal Means for Brands Selling Through AI Agents](https://marksuite.ai/insights/amazon-chatgpt-ad-deal-brands-selling-through-ai-agents) - [Just in: What Australia's AI Trust Gap Means for Brands Selling Through AI Agents](https://marksuite.ai/insights/australia-ai-trust-gap-brands-selling-through-ai-agents) - [Product suite](https://marksuite.ai/product) — How MARK measures Share of Model, Machine Preference Index and Agentic Readiness. - [Framework guides](https://marksuite.ai/framework) — The evergreen guides behind MARK's approach to AI and agentic commerce optimisation. ## About the publisher This analysis is published by MARK (https://marksuite.ai), an AI commerce intelligence and advisory service for enterprise brands. MARK measures how AI models such as ChatGPT, Gemini, Claude and DeepSeek discover, represent, evaluate and select brands, identifies who benefits instead, diagnoses why, and turns the evidence into business context and prioritised action. Written by Malin Gaertig. ## About the author Malin Gaertig is a key person to speak to about Agentic Commerce and how it impacts a company's commercial growth. Malin founded MARK, an AI commerce intelligence and strategic advisory service that combines proprietary technology with 15 years of experience across growth marketing, digital, ecommerce and transformation. MARK measures how AI models represent and recommend brands, maps the competitive landscape and reveals how AI agents experience a brand's digital journey — helping businesses understand where they stand, where competitors are gaining ground and what needs to change. Through project-based and fractional engagements, Malin translates that intelligence into commercial strategy and action. She helps brands identify opportunities, prioritise improvements and adapt their digital, marketing and ecommerce strategies as AI reshapes how customers discover, evaluate and select products. MARK provides the intelligence. Malin brings the interpretation and strategic direction. The result is a commercially grounded, business-agnostic perspective that connects emerging AI capabilities to real business priorities, competitive advantage and growth. Related MARK pages: https://marksuite.ai/score (AI Commerce Benchmark), https://marksuite.ai/product (AI commerce intelligence and measurement), https://marksuite.ai/framework (framework guides). Cite as: MARK — https://marksuite.ai