Australia's largest business lender is scaling AI through about 50 business-led AI squads, a single agentic and AI platform, and Ruby, an Anthropic-based assistant for more than 40,000 staff, while reporting task-level gains rather than a dollar value and cutting up to 2,500 roles it says were not driven by AI.
How is National Australia Bank (NAB) using AI?
National Australia Bank (NAB), Australia's largest business lender with A$1,145,896 million in total assets at 30 September 2026, reported its first year of scaled agentic AI with its FY26 results on 5 November 2026. It set up about 50 business-led AI squads working with partners including Anthropic and AWS, put a single agentic and AI platform in place, and gave more than 40,000 colleagues Ruby, an assistant built on an Anthropic platform. NAB reports task-level outcomes instead of a dollar value: a 20% reduction in the effort to prepare business credit memos, about 50% less preparation time for personal bankers, more than 90% less time for cloud cyber assessments, and selected legacy modernisation running three times faster than expected. Its Customer Brain engine draws on more than 3,500 AI models. Leadership was rebuilt around AI: Pete Steel, hired from Lloyds, became Group Executive Technology and AI on 1 October 2026, Mahya Knox joined from Commonwealth Bank as Chief AI Officer in January 2026, and an AI science team followed in April. In October 2026 NAB said it would cut up to 2,500 roles; CEO Andrew Irvine told Parliament the cuts were not driven by AI.
Lead bank
National Australia Bank Limited
Total assets (group, as reported, Sep 30, 2026)
A$1,145,896 million total assets[1]
Charter
Australian banking group; National Australia Bank Limited is an authorised deposit-taking institution supervised by the Australian Prudential Regulation Authority (APRA), with Bank of New Zealand (BNZ), the digital bank Ubank, JBWere (private wealth) and nabtrade
Headquarters
Melbourne, Australia
Flagship platform
Ruby, the Customer Brain and NAB's single agentic and AI platform
AI leadership
Andrew Irvine · Pete Steel · Mahya Knox · George Mathews · Shane Conway · Christian Nelissen · Jessica Cuthbertson · Lisa Marchant · Nitin Sachdeva · Steve Day · Nicholas Di Giuseppe · Sandro Bucchianeri · Connie Sokaris · Inder Singh · Patrick Wright
- Strategy: the 2026 annual report sets four AI areas (Banker AI, Customer AI, Process and engineering AI, Colleague AI) and says NAB deploys AI 'where it generates measurable business value and can be operated safely and responsibly within the Group's risk appetite'.
- Scale: about 50 AI squads working with global partners, a single agentic and AI platform, and almost 43,000 colleagues with access to approved AI tools (FY26). In the first half of FY26 NAB counted more than 7,000 engineers coding with AI tools and about 25,000 colleagues using AI for routine tasks.
- Outcomes, not a dollar figure: credit memo preparation effort down 20%, personal banker preparation time down an average of 50%, C&IB time to identify customer insights down 75%, cyber incident triage 80% faster and cloud cyber assessments down more than 90%. iTnews noted that, unlike its big-four rivals, NAB put no dollar value on its AI.
- Ruby: an Anthropic-based assistant for personal productivity, opened to more than 40,000 colleagues and revealed by CEO Andrew Irvine on the FY26 results call (November 2026).
- Customer Brain: launched in 2023 on Pegasystems' Customer Decision Hub; about 800 adaptive models in June 2024, more than 2,000 by June 2025 guiding over 50 million interactions a month, and 3,500 models across 90% of divisions by March 2026.
- Engineering: more than 6,000 engineers and technologists used Cursor or Amazon Q Developer by January 2026; Cursor trials showed a 40x leap in generating software requirements and a 5 to 6x uplift in development phases.
- Leadership: Pete Steel (from Lloyds) joined as Group Executive Digital, Data and AI on 19 November 2025 and took over technology as Group Executive Technology and AI on 1 October 2026; Mahya Knox (from CBA) is Chief AI Officer. iTnews reported in April 2026 that George Mathews (from QuantumBlack) would lead a new AI science team; NAB's June 2026 Genie release names Nitin Sachdeva as Head of AI Science.
- Governance: AI Risk became a standalone material risk in NAB's risk taxonomy in 2026; medium-risk use cases need peer review and high-risk ones an independent Model Validation review before go-live, and responsible-AI training is mandatory for all colleagues.
- Agents for customers: Patrick Wright, then group executive for technology, told Bloomberg in August 2026 that NAB would test the security and operational guardrails of an agentic AI platform, saying customers 'will expect one day to interact with us through agents'.
- Cloud: almost 90% of applications on cloud across AWS, Microsoft Azure and Google Cloud (October 2025), with the Knox data centre exited in 2026; NAB says the move cut critical and high outages by 89%.
- Workforce: NAB will cut up to 2,500 internal roles (about 6%) in FY27, with no material impact on customer-facing roles; iTnews, citing the Australian Financial Review, reported in November 2026 that about 1,200 of the cuts would come from technology and AI teams.
- Fraud and cyber: more than A$900 million invested in fraud, scams, cyber and financial crime protections in FY25, including AI; NAB Nexus, a 24/7 hub for AI-enabled threats, opened in June 2026.
What is Ruby, the Customer Brain and NAB's single agentic and AI platform?
NAB groups its AI work as AI@NAB: Banker.AI and Customer.AI for growth, Operations.AI, SDLC.AI, Obligations.AI and Cyber.AI for productivity, and Ruby.AI for colleagues, all on what its FY26 investor presentation calls a single agentic and AI platform. Ruby, revealed on the FY26 results call, gives more than 40,000 colleagues access to an assistant built on an Anthropic platform; the 2026 annual report says almost 43,000 colleagues can use approved tools including Claude, Copilot, Claude Code, Cursor, Devin and Gemini. The customer layer is the Customer Brain, a next-best-action engine on Pegasystems' Customer Decision Hub that NAB says draws on more than 3,500 AI models. Data sits on a Databricks lakehouse, with Databricks' Genie conversational analytics live since June 2026. About 90% of applications run in the cloud across AWS (primary for core applications), Microsoft Azure and Google Cloud. The 2026 annual report names AWS, Anthropic, OpenAI, Databricks, Google and Microsoft as its AI partners.[2][30][1][20][15][17]
Which AI and cloud companies does NAB work with?
NAB's 2026 annual report names six AI partners: Amazon Web Services, Anthropic, OpenAI, Databricks, Google and Microsoft, which it says provide the cloud infrastructure, AI models and platforms behind its AI ecosystem. Its staff assistant Ruby is built on an Anthropic platform, its Customer Brain runs on Pegasystems' Customer Decision Hub, and its engineers use Cursor and Amazon Q Developer. NAB runs about 90% of applications in the cloud across AWS, Microsoft Azure and Google Cloud.[1][30][20][19][17]
| Partner | What NAB uses | Status | Date |
|---|---|---|---|
| Anthropic | Platform behind Ruby, the assistant for more than 40,000 colleagues; programmes with Anthropic to upskill AI squads | In production | 2026-11-05[30][2] |
| Amazon Web Services | Primary cloud for core applications, Amazon Connect contact centre, Amazon Q Developer and Amazon Bedrock training | In production | 2024-09-19[18][17] |
| Microsoft | Azure cloud, relationship extended to the end of 2031; Copilot among approved AI tools | In production | 2025-10-20[17][1] |
| Google Cloud added as third cloud; Gemini among approved AI tools | In production | 2025-10-20[17][1] | |
| Databricks | Lakehouse data platform and Genie conversational analytics | Rolling out | 2026-06-18[15][21] |
| OpenAI | Named AI partner; large language models used in 2024 generative AI pilots | Partner | 2026-11-05[1][37] |
| Pegasystems | Customer Decision Hub behind the Customer Brain | In production | 2025-06-06[20][32] |
What results has NAB reported from AI?
NAB reports AI results as task-level improvements rather than a dollar value. Its FY26 results and 2026 annual report cite 20% less effort to prepare business credit memos, an average 50% cut in personal banker preparation time, 75% less time for corporate bankers to find customer insights, 80% faster cyber incident triage, more than 90% less time for cloud cyber assessments and selected legacy modernisation three times faster than expected.[2][1][30]
| Area | Reported outcome | Measured against | As of |
|---|---|---|---|
| Business lending credit memos | 20% average reduction in preparation time | Prior processing for users in scope | 2026-11-05[2][1] |
| Personal banker lead preparation | 50% reduction | Prior processing for users in scope | 2026-11-05[2][1] |
| C&IB customer insights | 75% reduction in time to identify insights | Prior processing for users in scope | 2026-11-05[2] |
| Design to code | 50% reduction in delivery time | Prior processing for users in scope | 2026-11-05[2] |
| Cyber | 80% faster incident triage; 90% less technology cyber assessment time | Prior processing for users in scope | 2026-11-05[2][1] |
| Transaction disputes (Customer Brain) | Customers kept informed are 30% less likely to call | Customers not proactively informed | 2026-06-18[15] |
What has NAB done on AI, and when?
- Nov 13, 2026
APRA plans AI stress tests for banks[41]
APRA's chair said the regulator intends to run stress tests of AI-driven risks from next year, after writing to banks in April about cyber threats from frontier AI models; its System Risk Outlook said frontier models from Anthropic and OpenAI are increasingly able to find and exploit software vulnerabilities.
- Nov 13, 2026
CEO tells Parliament the job cuts were not driven by AI[25]
Irvine told the House Economics Committee that NAB's workforce changes were not driven by AI, while acknowledging AI will ultimately have significant implications for the banking workforce.
- Nov 6, 2026
Ruby, built on Anthropic, opened to more than 40,000 colleagues[30]
Andrew Irvine revealed the staff assistant on the FY26 media call; iTnews reported that NAB, unlike its big-four rivals, put no dollar figure on its AI value.
- Nov 5, 2026
FY26 results: about 50 AI squads and a single agentic and AI platform[2][1][4]
NAB reported early outcomes from agentic AI tools (20% less time on business credit memos, 75% faster C&IB insights, 80% faster cyber triage) and named voice-enabled servicing agents, complaints automation and financial crime detection agents as the next wave in build. The annual report made AI Risk a standalone material risk.
- Oct 20, 2026
Up to 2,500 roles to be cut[24][40][39]
NAB said it would reduce its internal workforce by up to 2,500 roles, about 6%, in FY27, with up to 1,500 before the end of 2026 and a restructuring charge of about A$300 million; the release cited rapid technological change but did not mention AI, and the Australian Financial Review wrote that Irvine insists there is no link.
- Oct 9, 2026
Pete Steel at the parliamentary inquiry into AI[42]
Appearing with Westpac and ANZ, NAB's Group Executive Technology and AI told the inquiry that hosting data centres and AI training in Australia was a unique opportunity, and that it was extremely important for government fraud regimes to extend upstream.
- Aug 27, 2026
Founding partner of the Buy Australian AI Partnership[23]
NAB joined the federal government's programme connecting Australian AI startups with banks, insurers and superannuation funds through an eight-week accelerator.
- Aug 11, 2026
Technology and AI combined under Pete Steel[9]
With Patrick Wright and COO Les Matheson retiring from the Executive Leadership Team on 30 September, Steel became Group Executive Technology and AI and Shane Conway Group Executive Transformation and Operations.
- Aug 4, 2026
NAB to test guardrails for customer-facing AI agents[35]
Patrick Wright told Bloomberg NAB would soon test the security and operational guardrails of an agentic AI platform, with pilots over the next year, and already used agentic AI for some internal operations tasks.
- Jun 22, 2026
NAB Nexus opens against AI-enabled threats[16]
A 24/7 hub joining technology, cyber security, fraud, payments and physical security responders; NAB said it invested more than A$900 million in FY25 in fraud, scams, cyber and financial crime protections, including AI.
- Jun 18, 2026
Databricks Genie rolled out to the analytics community[15]
NAB said it was the first bank in Australia using the conversational analytics tool in its analytics community, saving 2 to 4 days of development per use case, with wider business access in pilot.
- May 4, 2026
Half-year results set out AI for growth, productivity and colleagues[5]
The 1H26 presentation reported a 44% increase in leads provided to bankers, more than 7,000 engineers coding with AI tools, about 25,000 colleagues using AI for routine tasks, 10 million contact-centre calls a year transcribed and summarised, and an AI leadership programme for 600 colleagues.
- Apr 28, 2026
AI science team announced[31]
Chief AI Officer Mahya Knox said a new AI science team, to be led by George Mathews from McKinsey's QuantumBlack, would build AI agent architecture, evaluation methodology and new AI products.
- Mar 20, 2026
Customer Brain grows to 3,500 models[32]
NAB's acting executive for customer decisioning said the engine drew on 3,500 AI models, was used across 90% of divisions and could suggest over 400 next best actions, and that NAB had started using Pega's GenAI Blueprint.
- Jan 20, 2026
- Jan 6, 2026
More than 6,000 engineers on Cursor or Amazon Q Developer[19]
NAB said Cursor trials produced a 40x leap in generating software requirements and a 5 to 6x uplift in development phases, and that over 6,000 engineers and technologists were using Cursor or Q Developer.
- Nov 19, 2025
- Nov 6, 2025
2025 annual report: AI Framework and over 100 initiatives[6]
The report says all AI initiatives are governed by NAB's AI Framework, aligned with the Australian Government AI Ethics Principles, and that over 100 AI initiatives were in pilot or production.
- Oct 20, 2025
Google Cloud added; almost 90% of applications on cloud[17]
NAB brought Google Cloud in alongside AWS and Microsoft Azure, extended its Microsoft relationship to the end of 2031, and said critical and high outages had fallen 89%.
- Sep 24, 2025
Mortgage file preparation cut from 48 minutes to three[34]
Group Executive Personal Banking Ana Marinkovic told an AWS event that an internal demonstration had cut preparation of a mortgage file from 48 minutes to three, and that NAB handles about 12 million calls a year on Amazon Connect.
- Jun 24, 2025
New Group Executive for Digital, Data and AI[8]
NAB named Pete Steel, Managing Director of Customer Engagement at Lloyds Banking Group and a former CBA Group Chief Digital Officer, to a new role leading digital, data and AI and the digital bank Ubank.
- Jun 6, 2025
Customer Brain reaches 2,000 adaptive models[20]
NAB said the engine, launched in 2023, guides more than 50 million interactions a month with over 2,000 adaptive models, 1,200 data points and 220 next best actions, and that term-deposit prompts had retained about A$92 million of deposits.
- Jan 13, 2025
CEO lists generative AI uses and 1,000 engineers on Amazon Q[28]
Andrew Irvine said NAB was using generative AI for banker knowledge management, marketing content, co-pilots that summarise reports and draft notes, and faster review of trust deeds, with more than 1,000 engineers using Amazon Q for coding.
- Nov 20, 2024
NAB joins BioCatch Trust Australia[26]
NAB joined an inter-bank network that shares behavioural intelligence in real time before a payment is made; it has used BioCatch's behavioural biometrics since early 2020 and invested in the company through NAB Ventures.
- Sep 19, 2024
AWS agreement extended, with Amazon Q Developer for 1,000 technologists[18]
NAB extended its long-term agreement with AWS to accelerate migration of critical workloads and develop generative AI, including automated code development; more than 1,000 technologists were trained and using Amazon Q Developer, and 84% of on-premises infrastructure had moved to a public multi-cloud environment.
- Jun 12, 2024
Customer Brain runs about 800 adaptive models[33]
NAB's customer analytics executive Jess Cuthbertson said the Pega-based Customer Brain drew on about 1,000 data attributes and about 800 adaptive models, covered three quarters of customer interactions and lifted engagement by 40%.
- May 27, 2024
Twenty generative AI use cases identified[37]
The Australian Financial Review reported that NAB had identified 20 generative AI use cases, including judging the seriousness of complaints, assessing trust deeds and tailoring customer messages, and was piloting many of them, including with large language models developed by OpenAI.
- Feb 7, 2024
Andrew Irvine named Group CEO[7]
The board appointed Irvine, then Group Executive Business and Private Banking and formerly head of Canadian business banking at BMO, to succeed Ross McEwan on 2 April 2024.
- Oct 23, 2023
NAB trials generative AI with Microsoft[38]
NAB said it was working with Microsoft on trials to embed AI into the software most bankers use to send correspondence, create presentations and record meetings, and was considering how widely to roll out generative AI for client emails and corporate credit reports.
Where does NAB use AI today?
| System | What it does | Status | Use case |
|---|---|---|---|
| Ruby staff assistant and approved AI tools[30][1] | Ruby, built on an Anthropic platform, gives more than 40,000 colleagues an assistant for personal productivity; almost 43,000 colleagues can use approved tools including Claude, Copilot, Claude Code, Cursor, Devin and Gemini for drafting, research, analysis and software development. | In production | Generative & agentic AI |
| Credit memo assistant for business bankers[1][2] | AI drafts and assembles business lending credit memorandums; for bankers using the tool, preparation effort fell 20%. | In production | Credit scoring & underwriting |
| Banker.AI preparation for personal and corporate bankers[1][2][5] | AI prepares personal bankers for customer conversations (preparation time down an average of 50%) and cuts the time C&IB bankers need to identify customer insights by 75%; leads provided to bankers rose 44% in 1H26. | In production | Generative & agentic AI |
| Customer Brain next-best-action engine[32][20][2] | Pega Customer Decision Hub engine drawing on more than 3,500 AI models, used across 90% of divisions, with over 400 next best actions such as dispute updates, home loan milestones and term-deposit prompts. | In production | Customer-facing chatbots |
| AI coding tools across the software lifecycle[19][5][2][1] | Cursor and Amazon Q Developer for more than 6,000 engineers by January 2026 and more than 7,000 engineers coding with AI tools in 1H26; design-to-code delivery time down 50%, and selected legacy modernisation three times faster than expected. | In production | AI-generated code & coding agents |
| Cyber.AI[1][2] | AI makes adoption of new cloud services safer and faster, cutting technology cyber assessment time by more than 90%, and speeds triage of cyber incidents by 80%. | In production | Cybersecurity |
| Databricks Genie conversational analytics[15] | Plain-English questions are turned into governed SQL queries with auditable answers; live in the analytics community since June 2026, saving 2 to 4 days of development per use case, with wider business access in pilot. | Rolling out | Generative & agentic AI |
| Obligations.AI monitoring[5][1] | AI-powered monitoring of customer obligations and commitments, piloted at Ubank in the first half of FY26 and described in the 2026 annual report as supporting risk management. | Pilot | AI governance (general) |
| Contact-centre AI on Amazon Connect[5][34] | About 10 million calls a year recorded, transcribed and summarised; Amazon Connect predicts why a customer is calling and routes the call, and AI agents handle simple chat queries among about 3 million chat requests a year. | In production | Customer-facing chatbots |
| Customer-facing agentic platform tests[35][2] | Testing of the security and operational guardrails of an agentic AI platform through which customers' agents could interact with NAB staff and systems; voice-enabled servicing agents and complaints case automation are in build. | Pilot | Generative & agentic AI |
| Behavioural biometrics, payment alerts and NAB Nexus[26][16][2] | BioCatch behavioural biometrics in digital channels since 2020, real-time payment alerts that led customers to abandon an average of A$470,000 of payments a day, and a 24/7 hub against AI-enabled threats; financial crime detection agents are in build. | In production | Fraud detection |
| Multi-cloud and AI partner set[17][1] | AWS (primary cloud for core applications), Microsoft (relationship extended to the end of 2031) and Google Cloud; AI partners named as AWS, Anthropic, OpenAI, Databricks, Google and Microsoft. | In production | Third-party & vendor AI |
| AI Framework and standalone AI Risk[1][6] | Mandatory responsible-AI training; risk-tiered approval with peer review for medium-risk and independent Model Validation for high-risk use cases; prohibitions on manipulative uses and social scoring; human involvement required for decisions with significant customer impact. | In production | AI governance (general) |
Use-case links open the tracker's rules-by-use-case matrix: which documents govern each system type.
What has NAB disclosed in numbers?
| Metric | Value | As of |
|---|---|---|
| Total assets | A$1,145,896 million[1] | Sep 30, 2026 |
| Colleagues with access to approved AI tools | Almost 43,000[1] | Sep 30, 2026 |
| Ruby staff assistant | More than 40,000 colleagues with access[30] | Nov 5, 2026 |
| AI squads | About 50 business-led, cross-divisional squads[2][30] | Nov 5, 2026 |
| Business credit memo preparation | 20% less effort for bankers using the tool[1][2] | Sep 30, 2026 |
| Personal banker preparation time | Down an average of 50%[1] | Sep 30, 2026 |
| C&IB time to identify customer insights | Down 75%[2] | Nov 5, 2026 |
| Cyber assessment and triage | Cloud cyber assessment time down more than 90%; cyber incident triage 80% faster[1][2] | Nov 5, 2026 |
| Customer Brain AI models | More than 3,500[2][32] | Nov 5, 2026 |
| Engineers coding with AI tools | More than 7,000[5] | Mar 31, 2026 |
| Leads provided to bankers | Up 44% in 1H26 against 1H25[5] | Mar 31, 2026 |
| Annual investment spend | A$1.8 billion, held broadly stable; productivity benefits of A$463 million[3] | Sep 30, 2026 |
| Applications in the cloud | About 90%[1][17] | Sep 30, 2026 |
| Fraud, scams, cyber and financial crime investment | More than A$900 million in FY25, including AI and advanced analytics[16] | Sep 30, 2025 |
| Scam and fraud losses prevented or recovered | More than A$385 million in 2025[27] | Dec 12, 2025 |
| Genie conversational analytics | 2 to 4 days of development saved per use case[15] | Jun 18, 2026 |
| Planned role reduction | Up to 2,500 internal roles (about 6%) in FY27; restructuring charge about A$300 million[24] | Oct 20, 2026 |
| AI initiatives in pilot or production | Over 100[6] | Sep 30, 2025 |
| Digital share of banking interactions | Around 98%; about 120 million NAB app and Internet Banking logins a month[29] | May 21, 2026 |
“Over 40,000 of our colleagues will have access to Ruby, and that's built on an Anthropic platform that we're very excited about.”[30]
Andrew Irvine · Group Chief Executive Officer · Nov 5, 2026
“While NAB's recent workforce changes were not driven by AI, we must acknowledge the reality that artificial intelligence will ultimately have significant implications for the future workforce in banking and in many other industries.”[25]
Andrew Irvine · Group Chief Executive Officer · Nov 13, 2026
“Customers will expect one day to interact with us through agents,”[35]
Patrick Wright · Group Executive, Technology and Enterprise Operations · Aug 4, 2026
“As we implement AI safely and at scale, we must have the right skills, capabilities and experience,”[31]
Mahya Knox · Chief AI Officer · Apr 28, 2026
“As it stands, our bankers can spend up to 50 per cent of their time on administrative tasks to set up or service accounts and loans for customers.”[14]
Pete Steel · Group Executive, Digital, Data and AI · May 29, 2026
“AI is going to do a lot of great things for how we live and how we work, but at the same time, AI is also going to help criminals and their threats much faster than when it was a human-only environment,”[16]
Andrew Irvine · Group Chief Executive Officer · Jun 22, 2026
“It's quite a unique opportunity to become relevant and a significant participant in AI globally,”[42]
Pete Steel · Group Executive, Technology and AI · Oct 9, 2026
“Everyone wants to jump into GenAI, but you still need good data to draw out the best insights,”[22]
Christian Nelissen · Chief Data and Analytics Officer · Aug 15, 2025
“In banking, we must balance speed with strong governance, and this approach lets us do both.”[15]
Jessica Cuthbertson · Executive, Data and Analytics · Jun 18, 2026
“This shift reduces our concentration risk, enabling us to seamlessly move critical services between platforms in real time.”[17]
Steve Day · Executive, Enterprise Technology · Oct 20, 2025
Who runs AI at NAB?
Pete Steel
Group Executive, Technology and AI (since 1 October 2026); joined as Group Executive, Digital, Data and AI on 19 November 2025 from Lloyds Banking Group[1][8][9]
Mahya Knox
Chief AI Officer (joined from Commonwealth Bank in January 2026; reports to Pete Steel)[36][13]
George Mathews
Named to lead NAB's new AI science team, per iTnews on 2026-04-28; joined from McKinsey's QuantumBlack, where he was senior principal data scientist[31]
Shane Conway
Group Executive, Transformation and Operations (since 1 October 2026); Group Executive, Transformation from 1 December 2025[9][10][1]
Jessica Cuthbertson
Executive, Data and Analytics (June 2026); previously Executive, Customer Decision and Data Science, leading the Customer Brain[15][20]
Lisa Marchant
Acting Executive, Customer Decisioning and Data Science (March 2026)[32]
Nitin Sachdeva
Head of AI Science, as named in NAB's Genie release of 2026-06-18[15]
Steve Day
Executive, Enterprise Technology (CTO Enterprise Technology), leading the multi-cloud programme[17]
Nicholas Di Giuseppe
Executive, Transformation Engineering (led the Cursor trials)[19]
Sandro Bucchianeri
Chief Security Officer[16]
Which regulators govern NAB's AI?
| Authority | Why it matters here | Documents |
|---|---|---|
| APRA / ASIC | APRA supervises NAB as an authorised deposit-taking institution and ASIC supervises its conduct. NAB's 2026 annual report makes AI Risk a standalone material risk with independent model validation for high-risk use cases. APRA wrote to banks in April 2026 about cyber threats from frontier AI models and in November 2026 said it will run AI stress tests from 2027, as NAB rolls out an Anthropic-based assistant and tests customer-facing agents. | APRA AI Letter to Industry (April 2026) APRA System Risk Outlook — November 2026 APRA CPS 230 |
| Reserve Bank of New Zealand | NAB's subsidiary Bank of New Zealand, the largest business lender in New Zealand, is supervised by the RBNZ, so AI used in BNZ's banking falls under the New Zealand prudential regime as well as APRA's group oversight. | |
| FSB | A systemically important Australian bank depending on a few frontier-model and cloud providers (Anthropic, AWS, Microsoft, Google) sits inside the FSB's AI monitoring and sound-practices work on third-party concentration. | FSB AI sound practices (final report, Oct 2026) |
| Basel Committee | A Customer Brain drawing on more than 3,500 models fed from a central data platform engages the BCBS 239 risk-data aggregation principles that apply to large banks. | BCBS 239 |
What does the public record suggest about NAB's AI strategy?
Analysis by BankingNewsAI from the sources cited on this page. Observations, not advice.
01
Outcomes are published; value is not
CBA reports about A$200 million of measured AI benefits for FY26 and DBS about SGD 1 billion for 2025. NAB publishes percentage improvements per task (20% on credit memos, 50% on banker preparation, 90% on cyber assessments) and A$463 million of total productivity benefits without attributing a share to AI. A value figure would make its AI spend comparable with its rivals.
02
The job cuts and the AI build are being read together
NAB says its 2,500 role cuts were not driven by AI, but the announcement came in the same quarter as Ruby, 50 AI squads and a Chief AI Officer, and iTnews, citing the AFR, reported about 1,200 cuts in technology and AI teams. Disclosing how AI changes headcount plans, as the annual report starts to do with skills pathways, would answer the question unions and analysts keep asking.
03
Business banking is where the AI is aimed
The most specific outcomes (credit memos, C&IB insights, banker preparation, leads) sit in business and relationship banking, which NAB says is about 75% of underlying profit. That makes business lending speed and share, rather than a group-wide AI metric, the place to look for returns.
04
A three-cloud, multi-model stance
NAB runs on AWS, Microsoft and Google clouds and names six AI partners while giving staff Claude, Copilot, Cursor, Devin and Gemini. That reduces dependence on any single provider, which NAB's CTO framed as lowering concentration risk, but it also multiplies the third-party models its AI Framework must evaluate.
Which leader rankings does NAB appear in?
30 of the computed leaders rankings include NAB, each ranked from the same sourced records (bank pages, filings, hiring snapshots, the agent tracker, the incidents register).
How does NAB compare with its peers?
- How does National Australia Bank compare with its peers on AI and technology?
- National Australia Bank competitors and alternatives, by business line
Its peer group on this site, by business model and size:
Who leads AI at NAB?
Pete Steel, Group Executive Technology and AI since 1 October 2026, owns NAB's technology and AI agenda; he joined from Lloyds on 19 November 2025 as Group Executive Digital, Data and AI. Mahya Knox, hired from Commonwealth Bank in January 2026, is Chief AI Officer and reports to Steel, iTnews reported in April 2026 that George Mathews would lead a new AI science team, and NAB's June 2026 release names Nitin Sachdeva as Head of AI Science. Group CEO Andrew Irvine presents the AI programme at results.
What is Ruby at NAB?
Ruby is NAB's internal AI assistant, built on an Anthropic platform and made available to more than 40,000 colleagues for personal productivity. Group CEO Andrew Irvine revealed it on the FY26 results media call in November 2026. NAB's annual report says almost 43,000 colleagues can use approved AI tools including Claude, Copilot, Claude Code, Cursor, Devin and Gemini.
How does NAB use AI?
NAB uses AI in four areas set out in its 2026 annual report: banker tools (credit memo drafting, conversation preparation), customer experiences (the Customer Brain next-best-action engine), process and engineering (coding tools, operations, obligations monitoring, cyber) and colleague productivity (Ruby and approved AI tools). About 50 AI squads work on priority use cases on a single agentic and AI platform.
What is NAB's Customer Brain?
The Customer Brain is NAB's AI-powered engagement engine, launched in 2023 on Pegasystems' Customer Decision Hub. It chooses the next best action for each customer across the app, bankers and contact centres. NAB says it draws on more than 3,500 AI models, is used across 90% of divisions and guided more than 50 million interactions a month by June 2025.
Is NAB cutting jobs because of AI?
NAB says no. It announced in October 2026 that it would cut up to 2,500 roles, about 6% of its workforce, citing simplification, competition and rapid technological change, and CEO Andrew Irvine told Parliament in November that the changes were not driven by AI. iTnews, citing the Australian Financial Review, reported that about 1,200 of the cuts would come from technology and AI teams.
Which AI companies does NAB work with?
NAB's 2026 annual report names Amazon Web Services, Anthropic, OpenAI, Databricks, Google and Microsoft as its AI partners. Ruby is built on Anthropic, the Customer Brain runs on Pegasystems, analytics use Databricks Genie, and engineers use Cursor and Amazon Q Developer. AWS is its primary cloud for core applications, alongside Microsoft Azure and Google Cloud.
Does NAB use agentic AI?
Yes. NAB's FY26 results report early outcomes from agentic AI tools, including 20% less time preparing business credit memos and 80% faster cyber incident triage, with voice-enabled servicing agents, complaints automation and financial crime detection agents in build. In August 2026 it said it would test guardrails for an agentic platform that would let customers' agents deal with the bank.
How does NAB govern AI?
Under its AI Framework, aligned with Australia's AI Ethics Principles, every AI use case is risk-assessed before go-live: medium-risk cases need peer review and high-risk cases independent Model Validation. AI Risk became a standalone material risk in 2026, responsible-AI training is mandatory, and decisions with significant customer impact require human involvement.
- 2026 Annual Report
- 2026 Full Year Results Investor Presentation
- 2026 Full Year Results Management Discussion and Analysis
- NAB announces its 2026 Full Year Results
- 2026 Half Year Results Investor Presentation
- 2025 Annual Report
- Andrew Irvine appointed CEO of NAB
- NAB appoints Group Executive for Digital, Data and Artificial Intelligence
- Executive Leadership Team updates
- NAB announces Executive changes
- NAB appoints new Group Chief Risk Officer
- Executive Leadership team appointment
- Growth in AI jobs offers window into careers of the future at NAB
- Australian economy getting real about AI
- NAB first bank in Australia rolling out conversational AI data tool to speed up customer insights
- NAB Nexus keeping customers safe in an AI driven threat environment
- NAB expands cloud-first approach to boost customer innovation
- Customers to benefit from NAB collaboration with AWS
- The tools and training transforming NAB's workforce to become 'the person with AI'
- The Brain Behind Better Banking: How NAB's AI is Making Banking More Human
- From Curious to Confident: NAB Dives into 40+ Hours of Data & AI Learning this August
- A call to think bigger about Data & AI
- NAB backs Australian AI innovators
- NAB simplifying operating model and improving alignment to customer priorities
- CEO Opening statement - House Economics Committee November 2026
- NAB joins BioCatch Trust Australia to protect customers from scams and fraud
- NAB Chair and CEO deliver 2025 AGM update to shareholders
- 'It's called the lucky country for a reason' NAB CEO Andrew Irvine on housing, AI and productivity
- Redesigned NAB app makes banking simpler and faster for customers
- NAB deploys AI for staff called 'Ruby' that is built on Anthropic
- NAB is building an "AI science" team
- NAB readies to scale AI-based customer interaction
- NAB says 800 adaptive models drive its 'customer brain'
- NAB wants more customers onboard with AI
- NAB tests AI agents in plan to deploy tech to banking customers
- NAB hires Mahya Knox from CBA as talent war heats up
- National Australia Bank is rolling out generative AI in these three areas
- NAB says AI will cut the drudgery for its bankers
- NAB redundancies and other bank job cuts could be the beginning of an AI-led workforce shift
- NAB to slash up to 2,500 jobs in massive restructure
- Regulator warns of 'interruptions' as AI models developed by Anthropic, OpenAI and others evolve
- Westpac, ANZ, NAB appear before parliamentary inquiry into AI — as it happened
When NAB 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