Every AI interaction is a marker of trust in financial services

Banks are putting AI agents in front of customers for payments queries, identity checks and fraud alerts. A communications platform executive argues that the reliability of what sits beneath those agents matters as much as the guardrails placed on them.

Alejandro Murcia is Director of Specialized Sales at Sinch, the customer communications platform. In this contributed piece he sets out his own view on how financial institutions can keep AI-powered customer interactions trustworthy once they are live.

Artificial intelligence is becoming an increasingly important part of how banks and financial institutions communicate with their customers. AI can make services faster, more accessible and more personalised, whether it is helping a customer understand a payment, complete an identity verification checkor respond to suspected fraud.

However, introducing AI into financial communications also creates a new challenge for an industry reliant on customer confidence. When an AI communications agent makes a mistake, the consequences can extend far beyond an inconvenient customer experience, potentially affecting someone’s ability to access their money, prevent fraud or receive the appropriate financial support.

Consequently, preventing these failures has become a priority for financial institutions. However, evidence suggests that investment in governance alone cannot guarantee reliable customer interactions.

Investing in governance is not enough

Financial institutions already recognise the importance of protecting customer interactions. Recent research from Sinch found that 78 per cent of financial institutions rank trust, security and compliance as a leading investment priority for their AI communications programmes. However, the sector’s 69 per cent rollback rate of AI communications shows that commitment and investment do not automatically translate into reliable performance.

Rollbacks are often a necessary response once an institution identifies that a live agent is behaving incorrectly. While the decision to withdraw an agent can demonstrate that monitoring and governance processes are working, identifying a problem after deployment is different from preventing it from occurring. Many problems originate below the policy or model layer. For example, customer data exposure, lost context and gaps in audit trails can occur within the communications infrastructure supporting an agent, while hallucinations and inappropriate responses may stem from the model, training data or prompting.

These different categories of failure require different solutions. Model-level guardrails are needed to govern what an agent can say and how it responds, while platform-native controls should provide capabilities such as personal data masking. Treating every problem as a model issue risks directing engineering time towards the symptoms while weaknesses in the wider communications architecture remain unresolved.

AI success is measured in production, not deployment

Once an AI agent is live, it is expected to perform during the moments when customers need immediate and dependable support. For a bank, this could be during an outage affecting mobile payments.

Failures during these periods can spread quickly across the organisation, as customers move towards contact centres and increase the workload for human customer service agents precisely when demand is already high. Engineering teams are then diverted towards diagnosing and rebuilding the agent, while inaccurate, delayed or insensitive interactions create wider reputational and regulatory risks.

Leaders should not treat deployment as the measure of AI success. A successful deployment is one that continues to perform reliably in production, with monitoring in place to identify when an agent goes wrong before the problem reaches the customer. This standard is particularly important in financial services, where institutions are responsible for protecting customers’ money, data and financial wellbeing. Meeting it requires both effective oversight and communications infrastructure that gives institutions clear visibility into how agents are performing, allowing them to identify and address problems before customers are affected.

Trust depends on the technology beneath the model

The infrastructure supporting an AI agent is, therefore, fundamental to maintaining trust once it begins communicating with customers. Sinch’s research found that satisfaction with communications infrastructure was a stronger predictor of success in AI deployment than investment levels, guardrail maturity or the length of time an organisation had been operating AI in production. Yet 42 per cent of organisations reported insufficient reliability for AI at scale from their current provider, while 37 per cent identified limited multichannel capabilities.

These shortcomings have specific implications for financial services, where maintaining context across a customer journey is essential. A customer discussing a suspicious payment through chat should not need to begin again when transferred to a fraud specialist by phone, while someone disclosing a vulnerable personal or financial situation should not repeatedly have to explain it because different systems cannot share the appropriate context securely.

More than half of the organisations surveyed are custom-building the ability to preserve context when customers move between channels because their platform does not provide it natively. As financial institutions introduce more agents across messaging, email and voice, continually recreating these capabilities can consume engineering resources and make governance more complex with every deployment.

The right platforms should instead provide a reliable foundation of native security, monitoring and orchestration capabilities, while financial institutions add safeguards reflecting their products, customers, risk profiles and regulatory obligations.

Building trustworthy AI communications at scale

AI presents financial institutions with an opportunity to make customer support more immediate, conversational and responsive, particularly in situations where speed and clarity can drastically affect someone’s financial wellbeing. Realising these benefits will require banks and financial institutions to look beyond the initial deployment and govern the complete system through which an AI agent receives information, communicates with customers and transfers responsibility to human teams.

The institutions that scale AI most successfully will be those that can keep their communications accurate, secure and connected as deployments become more complex. In an industry where every interaction can influence confidence, trustworthy AI will depend as much on the systems beneath the model as on the intelligence of the model itself.