What is Customers Bank's AI strategy?
Customers Bancorp treats AI as a business transformation run by its CEO rather than a technology project. At TD Cowen's conference on 17 November 2026, Sam Sidhu said 'I am essentially the de facto Chief AI Officer. We don't have a Chief AI Officer at the company', and that a December 2025 AI plan for 2027 was judged in January and February 2026 'just wasn't ambitious enough', so the goal moved from modules and productivity to '100% end-to-end automation'. He dated the start to January 2023, when the bank signed its enterprise agreement with OpenAI. On 27 April 2026 it expanded that into a multiyear collaboration with OpenAI engineers on site, aiming at lending, deposits and payments; Sidhu said '75% of our team members use tools powered by OpenAI'. On the Q3 2026 call (23 October) he said AI 'is moving from pilot to production': loan-underwriting cycle time down 95%, a rebuilt commercial deposit onboarding piloted in 10 weeks, and over 60 people focused on AI transformation. Results cited at TD Cowen: 76,000 hours saved, 800 agents created, a 47% efficiency ratio in Q3 2026 and a target of the low 40s by the end of 2027. No AI budget figure is disclosed.[1][2][3]
Data as of Nov 17, 2026 · Customers Bank profile updated Nov 1, 2026 · 29 sources cited on this page · 141 banks profiled on this site · Markdown
What has Customers Bank's CEO said about the AI strategy?
In date order, from the OpenAI release, the Q3 2026 call and the TD Cowen fireside chat.
| Item | What it says | Who | Date |
|---|---|---|---|
| OpenAI collaboration | 'We expect a fundamental re-engineering of how Customers Bank operates'; 'Today, 75% of our team members use tools powered by OpenAI' | Sam Sidhu, President and CEO | 27 April 2026[2] |
| AI-native goal | The collaboration should position the bank 'to become one of the first AI-enabled regional banks in the United States' | Customers Bank release | 27 April 2026[2] |
| Pilot to production | 'AI is moving from pilot to production, and it is already showing up in our numbers with more opportunity ahead' | Sam Sidhu, Q3 2026 call | 23 October 2026[3] |
| Top-down priorities | Loan underwriting cycle time 'down 95%'; commercial deposit onboarding rebuilt and piloted 'in just 10 weeks'; programmable subaccounts enabled on cubiX | Sam Sidhu, Q3 2026 call | 23 October 2026[3] |
| Plan rewritten | A December plan for 2027 'just wasn't ambitious enough'; the focus moved to '100% end-to-end automation' | Sam Sidhu, TD Cowen conference | 17 November 2026[1] |
| Who leads AI | 'I am essentially the de facto Chief AI Officer. We don't have a Chief AI Officer at the company, but I have CEO powers. I have the budget.' | Sam Sidhu, TD Cowen conference | 17 November 2026[1] |
| Business, not IT | 'The key essence is this is business transformation. It's not a technology change.' | Sam Sidhu, TD Cowen conference | 17 November 2026[1] |
| Governance | The deputy chief risk officer was seconded full time into the AI transformation, working with the CEO and general counsel on governance | Sam Sidhu, TD Cowen conference | 17 November 2026[1] |
What results does Customers Bank report from AI?
| Measure | Figure | Where | Date |
|---|---|---|---|
| Hours saved | 76,000 | TD Cowen conference summary | 17 November 2026[1] |
| Agents created | 800 | TD Cowen conference summary | 17 November 2026[1] |
| Loan underwriting cycle time | Down 95%, 'from three days down to 20 minutes', with a human in the loop | Sam Sidhu, TD Cowen conference | 17 November 2026[1][3] |
| Efficiency ratio | 47% in Q3 2026; target of the low 40s by the end of 2027 | TD Cowen conference summary | 17 November 2026[1] |
| Book close | A hard close in four days in Q3, against a usual five-day target | Sam Sidhu, TD Cowen conference | 17 November 2026[1] |
| Employee prompting | The average employee makes 8.5 prompts per day; prompts up 40% in the last 60 days | TD Cowen conference summary | 17 November 2026[1] |
What has Customers Bank not disclosed about its AI strategy?
Hours saved are self-reported and the agent counts include employee-built tools; at TD Cowen Sidhu said the bank tracks adoption, usage and self-reported savings, not only log-ins. No AI spending figure and no financial terms of the OpenAI collaboration are public.[1][2]
The bank has no chief AI officer; Sidhu says he holds that role in practice, and he said there is no playbook or peer bank to copy.[1]
What does the Customers Bank profile record on this topic?
The rows below are the entries of the Customers Bank AI strategy profile that bear on this page, as of Nov 1, 2026; the profile holds the full record of 32 dated moves, 17 use cases and 38 sources.
| Use case | What the record shows | Status | Source |
|---|---|---|---|
| AI-powered loan closing and account opening | Selected commercial loans closed in seven days versus 30 to 60; account-opening workflow cut 68%. | Pilot | [4][5] |
| AI KYC screening and OFAC false-positive clearing | Piloted in Q2 2026, per the second-quarter release, to make compliance processes more consistent and defensible; the January 2026 investor deck, citing AI use-case examples, reports AI-powered sanctions screening improved accuracy by 50%. Neither source describes the screening as in full production. | Pilot | [4][5] |
| AI call-centre scoring | 85% greater efficiency and full visibility into customer interactions. | In production | [5] |
| Employee-built agents and custom models on OpenAI | 500-plus agents; 75% of staff on OpenAI tools; on-site OpenAI engineering. | In production | [6][7] |
| Standardised AI risk framework and AI Innovation Lab | Enterprise governance for scaled deployment. | In production | [5] |
| AI charter, acceptable-use policy and use-case scoring | Monthly executive AI charter meetings on compliance, risk and policy; uses scored on value, cost, time to develop and risk. | In production | [8] |
| AI risk-factor disclosure | The 10-K carries a standalone risk factor on evolving AI regulation and names AI disruption of borrowers' industries and AI-enabled hacking as risks. | In production | [9] |
| AI-written software code | About half the bank's software code is written by AI, per the CEO. | In production | [10][11] |
| ElevenLabs voice and chat agents | Always-available service agents, real-time assist for relationship managers and conversational account opening. | Announced | [12] |
| Agentic commercial onboarding and cubiX agentic interface | Commercial deposit onboarding in minutes rather than hours; an agentic interface for cubiX payments targeting nearly 2 billion client API requests by year-end 2026. | Pilot | [13][4] |
| Qualia Shield wire-fraud verification for title and escrow | Wires originate and are received inside Qualia, with automated fraud verification on every transaction; Qualia is described as an AI-powered closing platform. | In production | [14] |
| AI fraud and deepfake detection (stated) | Customer guidance says the bank uses real-time fraud detection powered by AI, biometric authentication and deepfake and impersonation detection; no vendor, model or metric is disclosed. | In production | [15] |
| OpenAI, Microsoft and ElevenLabs relationships | Enterprise partnerships with OpenAI and Microsoft from 2023, OpenAI engineers on site since April 2026 and ElevenLabs agents from June 2026; Sidhu says the bank may co-create tools OpenAI could sell to other banks. | In production | [16][10][12] |
| AI across the three lines of defence | The 2026 shareholder letter aims to move risk, compliance and audit from sample-based to continuous monitoring with AI tools; the Q1 deck sets 'Reduce Risk' beside revenue and productivity. | Announced | [17][18] |
| cubiX MCP server and agentic-payment interface | A bank-built MCP server lets LLMs interact with payment rails; the bank says it was the first to publish an MCP, and targets nearly 2 billion client API requests through the agentic interface by year-end 2026. | In production | [19][20][13] |
| Agentic back-office and corporate functions | Deposit-product setup cut from an hour to a couple of minutes, legal documents reviewed in minutes, regulatory-filing accuracy tracking and a 60% shorter month-end close, all built in house. | In production | [20] |
| Palantir Foundry and AIP for deposit onboarding | Announced 15 October 2026: AI and automation for document collection, entity verification and risk-signal aggregation in a real-time digital application, built with Palantir engineers embedded beside the bank's AI team on its own data and systems. | Announced | [21][22] |
| Date | Move | Detail | Source |
|---|---|---|---|
| Oct 23, 2026 | Q3 2026 call: loan-underwriting cycle time down 95%, voice agents live | Sidhu says the scaled AI-powered loan underwriting process cut cycle time 95%, the rebuilt commercial deposit onboarding was piloted in 10 weeks (a more than 90% cycle-time reduction), programmable cubiX subaccounts were enabled for agentic-payment use cases, voice agents launched on the main customer service line and an AI digital twin pilot is expanding; over 60 people work on AI transformation and 12 and growing forward-deployed engineers from partners work in pods with in-house staff; the efficiency target remains a low-40s run rate by the end of 2027. | [23] |
| Oct 22, 2026 | Third-quarter 2026 results and deck: 76,000 hours, agent count doubled | The release says the team doubled the number of agents deployed in the quarter, cumulative estimated hours saved reached 76,000 (a nearly 65% increase, self-reported survey estimates as of 17 September 2026) and over $15 million of annualized capacity has been unlocked; a pilot of the new deposit onboarding process opened commercial deposit accounts in minutes rather than days, with the Palantir partnership expected to help scale it. Revenue per employee and pre-tax profit per employee are up 13% and 34% (annualized, 30 September 2026). | [24][25] |
| Oct 15, 2026 | Palantir collaboration on AI deposit onboarding | The bank will deploy Palantir's Foundry and AIP platforms to rebuild deposit onboarding; Palantir has embedded engineers alongside the bank's in-house AI engineering and technology teams, building on the bank's own data and systems. The two will seek to use AI and automation to streamline document collection, entity verification and risk-signal aggregation. The release cites deposits up 15% and client deposit accounts up more than 17% year over year; it gives no financial terms. | [21][22] |
| Aug 7, 2026 | Q2 2026 10-Q: no AI language | The quarterly report contains no reference to AI or OpenAI and reports no material change to the 2025 10-K risk factors; the Q1 10-Q (8 May) likewise has none. | [26][27] |
| Jul 24, 2026 | Q2 2026 call: underwriting, KYC and close-cycle results | A multi-agentic underwriting process closed loans within a week; AI KYC screening raised productivity 50%; month-end close fell 60%; 46,000 hours saved equals 24 FTEs; Sidhu leads a 40-person team, about 5% of the workforce; the bank says it was first to publish an MCP. | [20] |
| Jul 24, 2026 | Q2 2026 call: low-40s efficiency target for 2027 | OpenAI engineers on custom capabilities; multi-agent credit underwriting closed some C&I and CRE loans within a week; efficiency ratio targeted at a low-40% run-rate in 2027 from about 50%. | [28] |
| Jul 23, 2026 | Q2 2026 deck: 46,000 hours, 600 agents, 100% licensed | Agentic pods pair engineers with bankers; a 40-person AI mobilization team; top-down targets of a seven-day loan close, commercial onboarding in minutes, and a cubiX agentic interface handling nearly 2 billion client API requests by year-end 2026. | [13] |
| Jul 23, 2026 | Second-quarter 2026 results | Seven-day loan closings piloted; AI KYC screening and OFAC false-positive clearing piloted; net income to common $71.6 million. | [4] |
| Jun 29, 2026 | Qualia Shield wire-fraud verification goes live | Title and escrow customers can originate and receive wires inside Qualia, where the release says automated fraud verification runs on every transaction; Qualia is described as an AI-powered closing platform. | [14] |
| Jun 10, 2026 | cubiX page lists an MCP server and agentic-payment readiness | The bank's payments page offers a cubiX MCP server for LLM access to payment rails and describes the platform as ready for autonomous financial agents; Sidhu later says the bank believes it was the first to publish an MCP in 2025. | [19][20] |
| Jun 2, 2026 | ElevenLabs conversational agents | Voice and digital service agents, real-time assist for relationship managers and conversational account opening on ElevenAgents. | [12] |
| Apr 28, 2026 | 600 staff join an AI masterclass | Fortune reports the masterclass alongside the AI-clone call; the CEO says the aim is removing repetitive administrative work, not replacing judgment. | [29] |
Who leads AI at Customers Bank?
CEO Sam Sidhu. At TD Cowen's conference on 17 November 2026 he said he is 'essentially the de facto Chief AI Officer' and that the bank has no chief AI officer.
When did Customers Bank start using OpenAI?
Sidhu said the bank signed its enterprise agreement with OpenAI in January 2023; the April 2026 release says the relationship began in 2023 with ChatGPT Enterprise, when Customers Bank was an early adopter. The relationship was expanded into a multiyear collaboration with onsite OpenAI teams on 27 April 2026.
What is Customers Bank's efficiency target tied to AI?
A low-40s efficiency ratio by the end of 2027, from 47% in Q3 2026, per the TD Cowen conference on 17 November 2026. Sidhu described the aim as decoupling revenue growth from expense growth rather than adding simple productivity tools.
How many people work on AI at Customers Bank?
Over 60 people focused on AI transformation, per the Q3 2026 call on 23 October 2026, plus 12 and growing forward-deployed engineers from partners working in pods alongside in-house engineers. At TD Cowen in November Sidhu said nearly 30 in-house full-time engineers work on AI.
Does AI make credit decisions at Customers Bank?
Sidhu says no: bankers keep judgment and decisioning, while AI speeds document gathering, memo writing and financial analysis, per his TD Cowen remarks on 17 November 2026. Credit committees spend more time on risk discussion and less on administrative review, he said.
- Customers Bancorp at TD Cowen conference: ai push drives growth plan (with full transcript)
- Customers Bank partners with OpenAI on commercial banking
- Customers Bancorp (CUBI) Q3 2026 earnings call transcript
- Customers Bancorp Reports Results for Second Quarter 2026
- Customers Bancorp Q4 2025 Investor Presentation
- OpenAI will embed staff at Customers Bank under multiyear deal
- Customers Bank Announces Strategic Collaboration with OpenAI to Redefine the Commercial Banking Operating Model
- Customers Bank's Dynamic AI Strategy
- Customers Bancorp, Inc. Annual Report on Form 10-K for the fiscal year ended December 31, 2025
- This bank CEO let his AI clone handle an earnings call — now he's signing an OpenAI deal
- Customers Bank deepens ties with OpenAI
- Customers Bank and ElevenLabs to Deploy AI-Powered Conversational Agents Across Customer and Employee Experiences
- Customers Bancorp Q2 2026 Investor Presentation (Form 8-K, Exhibit 99.2)
- Customers Bank Deploys Qualia Shield for Customers, Advancing Wire Fraud Prevention for Title and Escrow Professionals
- Who's Responsible for Keeping Your Financial Information Safe?
- CUBI Q1 2026 Earnings Transcript
- Customers Bancorp Annual Shareholder Letter 2026 (Form 8-K, Exhibit 99.1)
- Customers Bancorp Q1 2026 Investor Presentation (Form 8-K, Exhibit 99.2)
- Digital Payments for Business (cubiX)
- Earnings call transcript: Customers Bancorp posts strong Q2 2026 growth
- Customers Bank and Palantir Collaborate to Build an AI-Native Future for Banking
- Customers Bank partners Palantir to enhance onboarding with AI
- Customers Bancorp (CUBI) Q3 2026 Earnings Call Transcript
- Customers Bancorp Reports Results for Third Quarter 2026
- Customers Bancorp Q3 2026 Investor Presentation (Form 8-K, Exhibit 99.2)
- Customers Bancorp, Inc. Quarterly Report on Form 10-Q for the quarter ended June 30, 2026
- Customers Bancorp, Inc. Quarterly Report on Form 10-Q for the quarter ended March 31, 2026
- Customers Bancorp Q2 Earnings Call Highlights
- CEO of a $25.9 billion bank had his AI clone lead the company’s earnings call—as Mark Zuckerberg builds his own digital twin
When Customers Bank moves on AI, you'll read it here first.
when one of the banks we cover moves on AI, it is in the next morning's brief · six sourced stories · 7 am PT · free
plus every regulation, bank and agent page update, the morning after · leave any morning