AI Agents for Financial Services: Client Onboarding and Education at Scale

Direct answer: An AI agent for financial services, and for AI wealth management in particular, handles client onboarding, product education, FAQ resolution, and appointment scheduling, within a clearly defined scope that excludes personalised investment advice. The face-and-voice interaction increases engagement and completion rates on what are typically high-dropout workflows (KYC, suitability questionnaires, product disclosure). EU AI Act Article 50 disclosure, GDPR consent, and MiFID II boundary controls are built into the deployment configuration, not retrofitted. Ojin supplies the face-and-voice layer for this, and the regulatory work sits in how the scope is configured rather than in which model is chosen.

Financial services is simultaneously one of the highest-value markets for conversational AI and one of the most tightly regulated. The client acquisition and onboarding funnel is expensive, complex, and full of friction points where clients abandon. KYC checks, suitability assessments, product disclosure documents, terms and conditions, each step loses some percentage of the potential client.

A face-and-voice AI agent that walks a client through these steps conversationally, explaining what each means, answering questions about why it is required, maintaining the professional register of a financial services institution throughout, changes the completion rate. The regulatory requirements stay exactly the same; what changes is whether the client experiences them as a guided process or a compliance obstacle.

What a financial services AI agent handles

The scope boundary matters more in financial services than in any other sector. A Human AI Agent is not a robo-advisor. It does not make personalised investment recommendations, assess suitability for specific products, or provide advice within the meaning of MiFID II, FCA, or SEC regulations. Providing that context clearly, to the client and in the system design, is the starting point.

Within that boundary, an Ojin Human Agent handles substantial value:

Client onboarding. Guiding the client through KYC data collection, identity verification prompts, beneficial ownership questions, and terms and conditions, conversationally, with explanations of why each step exists. Signicat's "Battle to Onboard" research, tracking 7,600 consumers across 14 European markets, puts digital onboarding abandonment at 68%, up from 40% in 2016, with process length and excessive information requests the two most cited reasons. A face-and-voice guided experience reduces abandonment because clients are less likely to close a conversation than a form.

Product education. Explaining what a product is, an ISA, a SIPP, a structured deposit, an ETF, in plain language, conversationally, at the client's level of financial literacy. Most clients do not read product brochures or KIDs (Key Information Documents). A face-and-voice agent that explains the product, fields questions, and confirms understanding before escalating to a human adviser is doing what KID disclosure is supposed to do, but rarely achieves.

FAQ and service queries. Account balance, transaction history, dividend timing, fee breakdowns, fund performance, all answerable from a verified data source without requiring a call centre agent.

Appointment scheduling. Connecting a client with the right human adviser, by specialism, availability, and the client's expressed need, at the moment the client is ready to engage.

The advice/information boundary described above is not a theoretical concern. The FCA's own July 2026 review into AI and retail financial services warns explicitly that personalised or continuous, adaptive AI recommendations can start to resemble regulated advice even when a firm intended the tool to provide general information only, and recommends the FCA examine within months whether its regulatory perimeter needs to be redrawn to keep pace. That is precisely why a financial services AI agent has to be configured, deliberately and visibly, to stop short of that line rather than assume it will not be crossed by accident.

The compliance architecture

Four compliance requirements apply to every financial services Human AI Agent deployment in the EU:

EU AI Act Article 50. Article 50 requires that people are told they are interacting with an AI system unless that is obvious from the context, and it applies from August 2026. On Ojin this is written into the agent's opening script rather than left to a page footer, because the requirement attaches to the conversation and not to the website around it. The disclosure is built into the opening statement: "[Bank Name] AI assistant, I'm an AI, here to help with your account and product questions. I'm not a financial adviser."

GDPR. Client data collected during the interaction, including name, account references, and any financial details shared, is personal data under GDPR. Data must be processed under a lawful basis (typically contract performance or legitimate interest for account-related queries), stored within the EU for EU clients, and subject to the firm's data retention policy.

FCA Consumer Duty (UK). The Consumer Duty holds firms to consumer understanding and consumer support outcomes, which is a useful design brief for an agent: it has to leave the customer able to act on what they were told, and it has to make getting help easy rather than harder.

MiFID II / FCA advice boundary. The agent must be clearly configured not to cross into personalised advice. This means: no product suitability recommendations for a specific client's circumstances, no "you should invest in X" outputs, explicit escalation to a human adviser when advice is the appropriate next step.

DORA (EU). The Digital Operational Resilience Act, Regulation (EU) 2022/2554, entered into force in 2023 and became applicable from January 2025, and requires financial institutions to document the ICT providers in their supply chain, including contractual provisions on data location, incident response, and exit arrangements for any provider supporting a critical or important function. Ojin, the Human AI Company, is a Berlin business running EU-based infrastructure, and its enterprise documentation covers the DORA third-party supplier requirements a bank's procurement team will ask about.

Record-keeping, the requirement firms forget until audit

Regulated firms are already required to keep records of client communications. The FCA's COBS 11.8 rules on telephone and electronic communication recording, with MiFID II sitting behind them, require firms to record communications relating to the reception, transmission and execution of orders, and to keep those records available on request. A face-and-voice agent holding a conversation with a client is generating exactly that kind of record.

The practical implication runs in the firm's favour. Unlike a branch conversation or a corridor phone call, every agent interaction is captured completely and identically, timestamped, and searchable, which is a better evidential position than the human channel it sits alongside. What has to be decided deliberately is retention, access, and where those recordings live, and that decision belongs in the deployment design rather than in a remediation project two years later.

Investor relations and institutional applications

Beyond retail banking and wealth management, Human AI Agents have an emerging application in investor relations. A listed company's investor relations page typically receives high-intent traffic, analysts, institutional investors, and retail shareholders looking for specific financial data, earnings commentary, and strategic context.

A face-and-voice agent that answers factual investor questions from a verified data set (annual reports, earnings transcripts, press releases), while clearly scoping out forward-looking guidance to a human IR contact, serves the institutional investor experience better than a static FAQ page. It is the IR director's presence, extended to every page visit. This is the register Ojin's Presence model exists for, the more lifelike of its two face models, where the face is standing in for a named executive and will be watched closely.

Regulated buyers still want to see the thing before the compliance conversation opens. The Ojin platform is self-serve, so a team can build a scoped, non-advisory agent and put it in front of its own risk function early, which is a faster route to an approved design than a specification written in the abstract. See what a self-serve AI platform is.

Frequently asked questions

Where does Ojin sit in this deployment?

As the technology provider and, under GDPR, the data processor. The regulated firm remains the controller and the authorised entity. Ojin runs EU-based infrastructure from Berlin, which is usually the first question a bank's procurement team asks.

Does an AI agent in financial services need FCA or BaFin authorisation?

An AI agent handling general information, FAQ responses, and account service queries does not require financial adviser authorisation in most jurisdictions. Crossing into personalised investment advice or product suitability recommendations does trigger authorisation requirements. The agent is designed to avoid that boundary, and escalates when a client's question moves towards advice.

How does the agent handle a client who is distressed about their financial situation?

The agent acknowledges the concern and routes to a human. Financial distress, account hardship, and potential debt crisis situations require human judgement and regulatory safeguarding. The agent is configured with escalation triggers for these signals.

Can the agent access live account data during the conversation?

With API integration to the core banking system or CRM, yes. The agent can query balance, transaction history, and account status from authenticated client sessions. Authentication requirements and data access scope are documented at docs.ojin.ai.

For the compliance guardrails that apply across regulated deployments generally, see brand safety when deploying a Human AI Agent and GDPR and AI agent deployments. For the category itself, see what a Human AI Agent is and how it actually works, what a Human AI Agent is and when it beats recorded video, and what a conversational AI agent is and how it differs from a chatbot. Wealth management, KYC and digital onboarding, investor relations, the robo-advisor comparison, and the SEC and FINRA detail each get their own article in this vertical. Across the rest of this batch, the closest neighbours are insurance, the neighbouring regulated vertical, and legal and professional services, a professional-services intake with the same advice boundary.

Ojin financial services enterprise deployment: ojin.ai/enterprise. Try a live Ojin agent: docs.ojin.ai.