Financial institutions are increasingly adding agentic AI to their broader generative AI projects. While generative AI models respond to prompts, agentic AI acts proactively, making their own decisions about how to achieve goals such as boosting efficiency or enhancing customer engagement.
This transition does not mean one AI model is replacing the other. As Chad Hamblin, director of global financial services at Microsoft, puts it, “generative and agentic AI are not competitive; they’re complementary. Agentic AI is an evolution of generative AI but they are very intertwined. The benefits of generative AI are still valid today; what we’re discussing now is the natural progression into agent-based capabilities. The fundamental difference is that agentic AI works in a much more proactive manner.”
Chad Hamblin is director of global financial services at Microsoft
With agentic systems, banks can deploy a set of specialised digital workers that tackle repetitive and high-volume tasks humans neither enjoy nor have time for. These agents operate 24/7, orchestrating multi-step processes, monitoring risk, triaging customer needs and ensuring information flows seamlessly across departments. The result is a banking organisation where bankers can focus on higher-value, advisory work.
Customer-facing bankers are an essential touchpoint in delivering a high-quality customer experience, but their ability to provide personalised service may be compromised by the need to take care of other tasks, such as managing workload, completing documentation and handling compliance checks. Agentic AI can step in as an added layer of support – a “wingman”, according to Hamblin, that completes tasks behind the scenes.
One use case is customer issue triaging within the branch. With an AI agent, customers can relay their needs, and the agent can automatically route them to the right banker, schedule appointments and ensure that specialists are available at the correct time. “It’s like a virtual concierge,” says Hamblin. “Agents gather customer information and nudge them towards the right people. Agentic AI can even help the branch organise and optimise their staffing schedule.”
VeriPark’s VeriBranch solution is focused on harnessing the power of AI to reduce time-consuming tasks, streamline branch operations and enable high-value interactions.
Edward Jones, for instance, has rolled out its Practice Assistant to its 46,000 employees working across 20,000 branches in the USA. The platform provides access to an “omnipresent assistant”, according to the firm’s general partner and head of EJ Labs Adrian Crockett, that provides access to various subagents but means that employees can come to one place to get the answers about their practice.
Employees across 20,000 Edward Jones branches are using agentic AI to better serve their customers
Agentic AI also shines with back-office tasks, like onboarding. Know-your-customer (KYC), identity verification, document collection and cross-departmental approvals are notoriously time-consuming. “KYC and onboarding are prime candidates for agentic AI to improve operational efficiency,” says Hamblin. “They involve extensive documentation and multiple cycles of verification, validation and follow-up.”
An AI agent can handle missing information, track document status, alert relevant departments and manage the entire workflow, reducing friction for both the banker and the customer.
For wealth managers, private bankers and personal finance advisers, agentic AI can constantly scan the market, analyse changes and prepare tailored recommendations. This doesn’t replace human judgment; it simply expands a banker’s capacity. “It would be like an investment assistant,” says Hamblin. “The agent wouldn’t make arbitrary decisions but would surface insights so wealth managers can better support a larger set of customers.”
Backbase’s AI-powered Financial Coach is a good example of this. The coach provides personalised insights, interactive simulators to explore ‘what-if’ scenarios and actionable, goal-based plans tailored to individual situations, ensuring customers are guided through every stage of their financial journey.
Similarly, agents can provide continuous monitoring for life events – such as the arrival of a new child, buying a home or saving for college – and recommend relevant banking products.
Microsoft partners are also playing a key role in enabling this level of personalisation. Personetics, for example, is using predictive analytics to analyse financial data in real time and anticipate customer needs, while Zafin’s platform streamlines product and pricing management, enabling tailored offers and quicker launches for new products like savings plans or rewards programmes.
The opportunities are nearly endless. “For example, agentic AI can help with churn-risk identification,” says Hamblin. “It can constantly look for triggers indicating potential churn and instantly reach out to customers to triage the situation. Human bankers can then step in to continue the conversation, but the early workload is handled automatically via AI Agents.”
In addition, agentic AI supports bankers in navigating compliance-heavy environments. Instead of searching for legal clauses or regulatory updates during a conversation, a compliance agent surfaces the right information or documents at the right moment. Research tasks – such as analysing corporate entities, financial filings or market events – can also be automated, letting bankers focus on insights rather than data collection. “Research takes time,” explains Hamblin. “Agents can guide bankers through it, pulling information together so they don’t have to do the heavy lifting.”
Even seemingly simple self-service use cases like answering FAQs and escalating issues can reduce friction for both customers and bankers. Unlike traditional chatbots, which are reactive, agentic AI can proactively surface information when customers might need it, such as during market changes or product updates.
Wells Fargo developed an agent using Microsoft Copilot Studio and Teams to help its 35,000 retail bank employees across 4,000 branches navigate procedures, regulations and complex banking systems. The agent provides instant access to guidance on 1,700 internal procedures, so employees can quickly locate the information they need without having to get support from a colleague. Now, 75 per cent of searches happen through the agent, cutting response times to customers from 10 minutes to just 30 seconds.
Wells Fargo employees are using an AI agent for quick answers about procedures and regulations
Lending is one of the most critical banking processes, involving multiple departments, strict compliance and sensitive financial data. A lot of the process consists of back-and-forth communication with customers, document intake and verification steps, complex financial modelling and risk assessments, all ideal candidates for agentic AI intervention. Platforms like Marble, a lending-as-a-service solution by Microsoft partner Tesselate, show how technology can streamline these processes. By combining third-party software, interoperability and managed services, Marble simplifies corporate loan servicing, enhances operational efficiency and enables institutions to respond quickly to market changes.
A good example is dynamic repricing. “What if you could have an AI agent dynamically reprice lending offers based on changes in the market?” asks Hamblin. “Instead of relying solely on rigid, rule-based processes, an AI agent could adjust offers in real time, presenting bankers with optimised recommendations for review.”
Another major use case is guiding customers through loan applications. Borrowers often struggle to understand the required documentation, the underwriting process and the steps involved, whether they are applying for a mortgage or a car loan.
“It’s a labour-intensive process with a lot of back and forth,” says Hamblin.
A team of AI agents can intake documents, verify income and identity, communicate with underwriting teams, check with legal and compliance and assemble materials for review, all while keeping the customer informed in real time. This reduces the uncertainty and frustration associated with loan applications, particularly for those who are unfamiliar with financial terminology.
Importantly, agentic AI does not replace human decision-making. “You always want the banker to make the final decision,” emphasises Hamblin. “There’s a difference between decision-making and guiding customers through a process.” Agents handle the orchestration, while bankers apply their expertise, knowledge and judgment.
Agentic AI can address deep-rooted operational challenges across the bank. Many institutions struggle with processes that span multiple systems, such as legacy platforms, disparate databases and customer service tools, which do not seamlessly communicate with each other. AI agents can be that connective tissue. For example, AI agents can detect back-office issues, identify mismatched data, flag stalled workflows and recommend resolutions. If a process requires multi-step coordination across departments like account changes or fraud investigations, an AI agent can orchestrate these tasks end-to-end.
At Scotiabank, reconciliation reports that once took up to 20 hours are now largely automated.
“Reports are painful and time-consuming,” said Matt Loos, manging director of global payments at Scotiabank, in a Microsoft-led session at Sibos 2025. “Now we have agents that clean the data, structure it so systems can talk to each other, and match data to create client insights reports. That allows us to scale something that was extremely manual.”
Diana Halder, partner for North American payments at EY, explained how banks can apply AI beyond customer-facing chatbots, which are already common across the industry. “Helping Scotiabank was about looking inside – adding value to the end user and upscaling the operations team,” she said. “The bank deployed AI in the back office, and that really makes it frontier.”
Scotiabank’s Matt Loos and EY’s Diana Halder (middle) presenting at Sibos 2025 with Microsoft’s Kathleen Woodward and Santhana Sankaramurthy
Compliance monitoring is another area where agentic AI can meaningfully reduce risk. “Bankers don’t have enough hours in the day to monitor everything,” says Hamblin. “An audit agent or AI-powered Microsoft Teams recording solutions like ASC Technologies’ Recording Insights can ensure checks are performed regularly, surface anomalies, generate suspicious activity reports and initiate follow-up tasks. This frees compliance teams to focus on the cases that genuinely require human review.”
The unifying impact of these operational use cases is that they allow banks to function more cohesively. Rather than siloed departments relying on manual processes, agentic AI ensures information circulates effortlessly across teams, improving efficiency and reducing errors.
As banks scale their use of AI, Microsoft identifies three evolving phases of agentic adoption. “The first phase is human with assistant where AI supports tasks, research and information gathering to improve individual productivity,” says Hamblin. “The second is human-led agents where AI orchestrates multi-step processes, but employees remain firmly in control. And the third is human-led, agent-operated systems where agents coordinate entire processes across the enterprise, with humans supervising, guiding and focusing on high-value decision-making.
“Agentic AI does not replace bankers, it empowers them and gives them more time, insight and capacity to deliver the experiences customers want: personalised guidance, faster service and greater financial confidence. The banks that orchestrate this harmony best will set the tempo for the next era of financial services.”
Partner perspectives
Selected Microsoft partners share how they are supporting financial services organisations to modernise their workplace and boost productivity.
“Altigen has been helping banks and credit unions enhance communications and customer interactions for more than 20 years,” says Mike Plumer, vice president of sales at Altigen Technologies. “Today, we work with over 1,500 institutions through our partnership with Fiserv. Our focus is on helping financial organisations retire outdated systems and consolidate everything within Microsoft 365.”
“Tesselate helps financial services organisations modernise the workplace through a phased, secure and governable adoption of agentic AI,” said Casimir Veisblat, managing director at Tesselate. “Our solution converts processes like letter of credit and guarantee issuance into adaptive, goal-driven workflows; enabling straight-through processing where permitted, and human-in-the-loop control where required. Using the Microsoft Power Platform, Copilot and the Azure AI Foundry ecosystem, we enable business users to build, orchestrate and monitor workflows across existing systems; typically reducing manual effort by 70 per cent in targeted areas and accelerating turnaround times through more consistent execution.”
“By seamlessly capturing and archiving every chat, call, meeting and Microsoft 365 Copilot interaction, compliance and supervision remain fully integrated – not obstacles to collaboration,” says Goutam Nadella, chief product officer at Smarsh. “The AI-driven Smarsh platform automates risk detection and analytics, turning everyday communications into meaningful intelligence.”
“VeriPark helps financial institutions modernise their workplace by combining Microsoft’s cloud, data and AI technologies with 25 years of deep financial industry expertise,” says Michel Diab, chief marketing and strategy officer at VeriPark. “Through Microsoft Azure, Dynamics 365, Copilot and Power Platform, we empower frontline and operations teams with AI-driven insights, automated workflows and secure collaboration.”
“Zafin’s AI-powered product and pricing platform modernises how banks design, price and deliver products decoupling innovation from legacy cores and accelerating time to market,” says Chris Dickin, head of strategic partnerships at Zafin. “By unifying product, pricing and customer data, Zafin applies AI for real-time insights, profitability simulation, dynamic segmentation and intelligent offer design.”
Discover more insights like this in the Winter 2025 issue of Technology Record. Don’t miss out – subscribe for free today and get future issues delivered straight to your inbox.