The financial sector is done experimenting with artificial intelligence. In 2026, the conversation has shifted from pilots to full-scale deployment — and the stakes have never been higher.
Banking and insurance executives no longer talk about AI as a novelty. They talk about it as infrastructure. The priority now is weaving generative AI directly into underwriting pipelines, trade accounting systems, client onboarding, claims processing, and customer service workflows — not as a supplemental layer, but as a core operational component.
This is a structural transformation. It demands clean, unified data, embedded oversight, and returns measured in dollars and decisions.
From assistants to agents
The first wave of generative AI in financial services was modest. It helped employees draft emails faster, summarize documents, and boost internal productivity. That era is largely over.
Saachin Bhatt, co-founder and COO at Brdge, captured the evolution plainly.
“An assistant helps you write faster. A copilot helps teams move faster. Agents run processes.”
That single distinction is now driving enterprise strategy across the sector. Financial institutions are constructing agentic AI systems capable of reading signals, making decisions, drafting communications, routing approval requests, and completing tasks — all within predefined compliance controls.
Bhatt describes this model as a “Moments Engine.” It detects real-time signals across digital touchpoints, applies decision logic, generates compliant messaging, routes tasks for human review when necessary, executes actions, and learns from every outcome.
Many banks already have pieces of this architecture in place. Few have connected them into a seamless, end-to-end system.
The obstacle is not the technology itself. It is the coordination challenge — legacy systems that don’t talk to each other, siloed data environments, and compliance frameworks that must be satisfied before anything can scale.
Governance moves to the core
In financial services, speed without trust is a liability. Governance is no longer an afterthought. It is a technical requirement baked into the design of AI systems from day one.
Jonathan Bowyer, former marketing director at Lloyds Banking Group, was direct about the regulatory exposure.
“Legitimate interest is interesting, but it’s also where a lot of companies could trip up,” he warned.
Institutions are responding by embedding compliance controls into prompt structures, data pipelines, and fine-tuning processes — catching problems before they reach the output stage rather than reviewing them after the fact.
Farhad Divecha, group CEO at Accuracast, argued that optimization must function as a closed loop — data insights feeding creative improvements, while quality assurance protects brand integrity at every step.
Customers must also know when they’re interacting with an automated system. Escalation pathways to human agents must remain clear and accessible.
This compliance-by-design approach allows institutions to scale automation responsibly — meeting regulatory obligations, such as Consumer Duty in the U.K., without sacrificing agility.
AIG reports measurable gains
American International Group has emerged as one of the clearest case studies in what full-scale AI integration can actually deliver.
At a recent investor presentation, AIG executives reported faster-than-expected results from embedding generative AI into underwriting and claims workflows. Chief executive Peter Zaffino, who had initially described early projections as “aspirational,” reversed course.
“We see the abilities are much greater,” he said.
Zaffino described the impact on submission processing as a genuine breakthrough.
“We’re seeing a massive change in our ability to process a submission flow — without additional human capital resources. That has been the biggest surprise.”
AIG’s internal platform, AIG Assist, is now live across most commercial lines. An orchestration layer coordinates multiple AI agents simultaneously within underwriting workflows.
Lexington Insurance, AIG’s excess and surplus division, has set a target of 500,000 annual submissions by 2030. It already surpassed 370,000 in 2025 — ahead of schedule — powered by generative models that extract and summarize incoming documentation at scale.
During the integration of Everest’s retail commercial business, AIG built a full ontology of the Everest portfolio and aligned it with its own systems, accelerating the renewal process dramatically. A comparable approach was used to launch Lloyd’s Syndicate 2479 in partnership with Amwins and Blackstone, with technology support from Palantir. Large language models assessed portfolio fit against risk appetite criteria in real time.
The results speak for themselves. This is no longer proof-of-concept territory.
Goldman extends AI to operations
Goldman Sachs is taking AI further. The firm is moving it beyond developer tools and into everyday business operations.
According to reports in American Banker, Goldman plans to deploy Anthropic’s Claude model in trade accounting and client onboarding. Chief information officer Marco Argenti made the case for why rule-based systems alone fall short. At scale, rigid automation leaves thousands of edge cases unresolved. Generative AI can apply contextual reasoning precisely where rules are ambiguous.
Goldman developers already use Claude alongside Cognition’s Devin agent to write and test code under human supervision. That hands-on experience is now informing the push into broader operational use.
Indranil Bandyopadhyay, principal analyst at Forrester, explained why generative AI fits reconciliation work so well. Trade accounting involves comparing fragmented data across ledgers, confirmations, and statements — a task that benefits directly from the large context windows modern AI models offer.
Claude operates at the workflow layer. Core systems of record retain their authority.
Jonathan Pelosi, head of financial services at Anthropic, noted that Claude is trained to flag uncertainty and provide source attribution — creating an audit trail that reduces hallucination risk and builds institutional confidence.
Argenti also pushed back on the notion that AI is inherently more vulnerable to manipulation than humans. Social engineering exploits human psychology, he argued. Automated systems can detect subtle anomalies at scale that humans might miss entirely.
The operational objective is clear: fewer exceptions, faster throughput, sustained oversight.
Data restraint and generative search
Personalization has evolved into something more sophisticated — anticipation.
Bowyer offered a sharp observation about shifting customer expectations.
“Customers now expect brands to know when not to speak to them as opposed to when to speak to them.”
That level of restraint requires truly unified data — connecting branches, apps, and contact centers so that a customer flagged as distressed in one channel is not blasted with promotional content in another. When that fails, trust erodes fast.
Search behavior is also changing in ways that matter to financial brands. Divecha noted that generative AI answers increasingly appear outside traditional websites.
“Digital PR and off-site SEO is returning to focus because generative AI answers are not confined to content pulled directly from a company’s website,” he said.
This has given rise to a new discipline: Generative Engine Optimization, or GEO. Financial institutions are structuring their public data specifically so that external AI systems cite them accurately — getting ahead of a shift in how customers discover and evaluate financial services.
The 2026 mandate
The directive for financial institutions in 2026 is unambiguous.
Unify your data. Hardcode your governance. Build agentic orchestration. Optimize your information for generative discovery.
AI is no longer a feature. It is embedded in the fabric of core financial operations. The competitive advantage going forward will not come from which model a firm uses — it will come from how well that firm integrates generative AI across its systems, workflows, and people.
The institutions that pair automation with sound human judgment are the ones setting the pace. Everyone else is catching up.
What’s your take on the financial sector’s shift to full-scale AI deployment? Are banks moving too fast — or not fast enough? Please share your thoughts in the comments below.