Popular searches
Not sure where to start?Talk to our team about running sovereign AI in your environment.
↑ ↓ Navigate↵ OpenEsc Close
Commonwealth Bank of Australia scaled AI across the bank with H2O.ai, cutting fraud 30% and customer scam losses 76% while training more than 900 analysts.
As Australia's largest bank, CBA protects millions of customers from fraud and increasingly sophisticated scams. Scam losses hit a peak in 2022, and stopping them meant detecting and acting on threats in real time, across the entire customer base.
As Australia's largest bank, Commonwealth Bank of Australia protects more than 16 million customers from fraud and increasingly sophisticated scams. Scam losses had climbed to a peak in 2022, and stopping them meant detecting and acting on threats in real time, at the scale of the whole customer base.
CBA built its customer engagement engine on H2O.ai, combining predictive and real-time generative AI for fraud detection, transaction abuse monitoring, and customer protection. The bank runs thousands of models over billions of data points in real time, and has created more than 50 generative AI use cases across the business, from Bills Sense to Benefits Finder and natural disaster support.
CBA reduced fraud by 30% and cut customer scam losses by 76% from their 2022 peak. The same AI foundation now powers personalized, proactive services for more than 16 million customers, keeping them safe in ways that were not possible before.
“From our peak in 2022, we've reduced customer scam losses by 76%. AI has allowed us to keep our customers safe in ways that simply weren't possible before.”Dan JermynChief Decision Scientist, CBA
To deliver AI at scale, CBA needed machine learning in the hands of data scientists, engineers, and business users across the organization, not just a central team. Building each new model took weeks, slowing how fast the bank could anticipate customer needs.
To deliver AI at scale, Commonwealth Bank needed machine learning in the hands of data scientists, engineers, and business users across the organization, not just a central team. Building each new model took weeks, which slowed how fast the bank could anticipate customer needs.
CBA formed an exclusive financial-services partnership with H2O.ai and adopted H2O AI Cloud across the organization, backed by a dedicated H2O.ai team including Kaggle Grandmasters and ML engineers. A no-code approach let employees far beyond the core data science team build highly predictive models quickly, with governance, documentation, and explainability built in.
CBA cut the time to prototype a new model from weeks to two days and trained more than 900 analysts on H2O.ai. Machine learning now runs across every business unit, and the bank operates as an AI-powered organization serving more than 16 million customers.
“We are building an AI-powered organisation for the benefit of our customers.”Dr. Andrew McMullanChief Data & Analytics Officer, CBA
CBA had transformed how it works with structured data, analyzing billions of transactions, but the insight locked in unstructured documents was largely untapped. With 16 million customers and hundreds of pages of documents each, traditional OCR could not keep up with the varying formats and poor-quality scans across onboarding, KYC, and credit assessment.
Traditional OCR and RPA could not keep up with the varying formats, templates, and poor-quality scans, especially for KYC, where colleagues spent significant time manually extracting and verifying information from passports, licenses, and birth certificates.
Through its partnership with H2O.ai, CBA got early access to H2O Document AI and stood it up as a group capability in about four months. The platform adds intelligence to OCR and RPA, combining classification models to identify document types, noise detection and reduction models to handle poor scans, and OCR with advanced NLP to read and structure the text.
CBA applied it first to KYC and ID extraction, then to credit-assessment documents like pay slips and bank statements, and to procurement, accounts payable, and accounts receivable, including three-way matching of purchase orders, invoices, and receipts.
CBA now processes millions of documents per day on the platform and has achieved 50 to 85% accuracy and automation across varying document types. The bank moved from automation to intelligent automation, unlocking insight from documents, text, and image data that legacy OCR could not reach.
Document AI opens faster customer onboarding, lower operational costs, and productivity gains across both customer and back-office processes.
“This really is just the beginning of where we can start to embed and radically reimagine our day-to-day operations and make lives better for our colleagues and our customers.”Sonal SuranaGeneral Manager, CBA
Commonwealth Bank of Australia is Australia's leading provider of integrated financial services, spanning retail, premium, business and institutional banking, superannuation, insurance, and share-broking. Founded in 1911 and headquartered in Sydney, CBA serves 15.9 million customers and employs around 48,900 people, with operations across New Zealand, Asia, the United Kingdom, North America, and the European Union.
More at https://www.commbank.com.au/about-us/our-company.html