BankingNewsAI Daily Brief ·
Google Cloud introduces the Gemini agent
Banking AI
Financial institutions & fintech technology
Chime saves 250,000 hours with AI call summaries
For the contact-center head at a regional bank
Chime sets a benchmark for call-summary automation: 250,000 hours and $700,000 a year. Your bank must decide whether its call records, retention rules and complaint process can support AI summaries before pursuing similar savings.
Chime deployed AI call summaries for agents on all calls after testing the application through 2023 and launching it in February 2024. It uses Amazon Transcribe to convert calls to text, redacts identifying elements, and sends transcripts to Claude models in Amazon Bedrock for summaries. Chime says the tool cuts average agent handling time by 18 seconds and saves more than 250,000 hours per year, or about $700,000 in annual efficiency gains. Agents use summaries for repeat callers, while Chime uses them to analyze calls and investigate anomalies. Chime says service net promoter score rose by five points.
→ Action
Contact-center operations: Test AI summaries against call recordings for accuracy, complaint reconstruction and retention requirements before deployment.
Read article → from Amazon
OnDeck automates small-business loan approvals
For the small-business lending head at a regional bank
OnDeck raises customer expectations for near-instant working-capital decisions. Your bank must decide whether speed is worth greater automated-decision risk, and prove that fair-lending outcomes and credit performance do not drift.
OnDeck launched automated final decisioning for small-business loan applications, extending automated approvals from eligible existing customers to new applicants across channels. Its AI-powered verification and machine learning can cut funding decisions from a few hours to under a minute. Qualified applicants can receive approval without stipulations and move directly to funding without additional documents. OnDeck says decisions remain subject to its existing underwriting standards and ongoing model-performance monitoring. The company expects to route a growing share of its billions in annual originations through automated decisioning.
→ Action
Model risk: Test automated small-business approval models for fair-lending outcomes and credit-performance drift by applicant segment.
Read article → from PR Newswire
nCino partners with Parlay to power AI-native SBA lending
For the SBA lending head at an nCino bank
The speed gain will expose any vague SBA credit policy. Set lender-owned rules and exception reviews for eligibility and deal structuring before automated intake sends more files toward underwriting.
nCino partnered with Parlay to integrate Parlay’s Loan Intelligence System with nCino Commercial and Small Business Solutions for SBA lending. The live integration has processed more than $500 million in SBA loan inquiries in 2026, including for First Internet Bank. A one-click handoff handles intake, eligibility and deal structuring before a banker opens the file. Parlay customers report that work to reach a credit-ready file fell by up to 80%. nCino will resell and support Parlay’s system for customers that prioritize SBA lending, while Parlay will continue to serve the broader SBA market directly.
→ Action
SBA credit policy: Set approval rules and exception review for Parlay eligibility and deal-structuring handoffs.
Read article → from Financialcontent
General AI
Large language models & AI infrastructure
Google Cloud introduces the Gemini agent
For the CIO of a regional bank
A single agent across staff tools and bank systems turns access control into a deployment gate. Do not let broad productivity goals outrun tool-level permissions, audit logs and hard spending limits.
Google Cloud introduced the Gemini agent, a universal agent for work, at Gemini at Work 2026. Google says users can start work in a Gemini prompt window for knowledge work, questions, content creation and coding. The agent plans work, uses skills and tools, connects to Cloud customers’ business systems, and returns finished work in documents, inboxes and developer environments. Google says Gemini chooses the best model for each job and includes cost controls, security, administration and governance for enterprise customers.
→ Action
Identity and access management: Define tool-level permissions, approval rules and audit-log requirements before connecting any agent to bank systems.
Read article → from Google
Anthropic commits $150 million to Genesis Mission
For the AI procurement head at a regional bank
Anthropic’s federal commitment makes Claude capacity, support and data terms a supplier-management issue, not just a model-choice issue. Your contracts should state what enterprise service levels and data boundaries survive public-sector demand.
Anthropic committed $150 million over three years to the federal Genesis Mission, which uses AI to accelerate scientific and technological discovery. The funding will make Claude available to more than 15 agencies, including NASA, the National Institutes of Health and the National Science Foundation. Anthropic will provide Claude, Claude Code and API credits to several hundred research projects. It will also provide training, onboarding and technical support, while working with agencies and national laboratories on fusion energy, quantum computing and other administration science priorities.
→ Action
Vendor management: Ask Anthropic whether Genesis Mission allocations alter Claude enterprise support, capacity commitments or data handling terms.
Read article → from Anthropic
Get this in your inbox every morning
Free · No spam · Unsubscribe anytime