Customer-facing AI in financial services has quietly become one of the industry’s main front doors, and AI chatbots in banking were the first wave. The US Consumer Financial Protection Bureau estimated that roughly 37% of the US population interacted with a bank’s chatbot in 2022, and that each of the ten largest commercial banks had deployed one. The newest generation goes further than answering balance questions. Voice and chat agents now explain insurance policies, walk customers through applications and complete purchases on the spot.
Jeen builds an enterprise AI operating layer, and its production customers include Direct Insurance, one of Israel’s leading digital insurance companies and a pioneer of insurance sold directly online and by phone. In finance, a conversation with a customer carries the same obligations as any other customer interaction. Jeen’s approach, built on runtime governance and full traceability, treats the AI holding that conversation accordingly from the first word.
The risks regulators have already named
The rules are taking shape through supervisors. In its June 2023 report on chatbots in consumer finance, the CFPB listed what can go wrong: inaccurate answers, failure to recognize when a customer is invoking their rights, “doom loops” that leave people stuck with no route to a human, privacy and security gaps, and failures to comply with federal consumer financial law.
European insurance supervisors have been equally specific about AI in insurance. In August 2025 the European Insurance and Occupational Pensions Authority published an Opinion on AI governance and risk management, setting out how existing insurance rules apply to AI. Its expectations cover data governance, record-keeping, fairness, cyber security, explainability and human oversight.
Read side by side, the two documents point to the same practical requirements. Answers have to come from verified information, every conversation has to leave a record, and a person has to be reachable when the AI is out of its depth.
Grounded answers and a clear way out
Jeen builds those requirements into the product. Jeen Talk, its voice and chat agent, understands customer intent, retrieves knowledge grounded in the company’s verified systems, takes action across business systems and escalates to human teams. It works across phone, website, email and WhatsApp, and connects to existing contact-center telephony instead of replacing it.
Grounding addresses the CFPB’s first concern directly. An agent that answers from a company’s approved policy documents and product data has far less room to improvise a wrong answer about coverage or fees. Intent recognition and escalation address the next two. A customer raising a dispute or a complaint can be passed to a person instead of being looped back to the start.
Alt text: Abstract illustration of a customer conversation handed from an automated channel to a human team
Every conversation on the record
Record-keeping appears by name in EIOPA’s expectations, and it sits at the core of conversational AI governance. The agent runs inside the same governed layer as everything else on the platform, combining runtime governance, enterprise security and full traceability. Policies and access controls apply to every conversation as it happens, and each one is recorded so a compliance team can review what the AI said and why.
When a supervisor or internal auditor asks how a particular answer was produced, the team can trace it back to the information the agent drew on.
The same layer brings FinOps for AI to customer service, so the cost of the agents handling customer conversations is tracked in real time, agent by agent. The platform is ISO 27001 and SOC 2 Type II certified and can run as SaaS, in a private cloud, on-premises or fully air-gapped, which matters to institutions whose customer data cannot leave their own infrastructure.
Results from insurance and banking
Jeen reports concrete outcomes from both sectors. In insurance, it built AI-guided journeys that explain policies in plain language, answer questions in real time and let customers complete a purchase by text or voice. It puts answer accuracy at 95%, and cost per closure on voice-led journeys at more than 60% below the previous level.
In banking, the work sat behind the customer conversation. Its team automated the review of international credit-network directives, turning lengthy documents into assigned business tasks with a full compliance record. It reports 95% document-processing accuracy and full prevention of missed-directive penalties. For institutions that answer to supervisors, accuracy and an audit trail carry as much weight as speed.
Why the governance question comes first
The company’s AI on Your Terms campaign, launched in July 2026, asks leaders to account for three things: the location of their AI, who has control over it, and whose interests it ultimately serves. Moti Krispil, the company’s Chief Strategy and Growth Officer, framed those as questions with a deadline. “The leaders who ask them now will be the ones still standing in five years,” he said.
For banks and insurers putting AI in front of customers, the second of those questions arrives first, often in the form of a supervisor’s request. Jeen built its operating layer so that request can be met from records that already exist.
