What proprietary AI has UBS built in-house?
UBS builds its own assistants, platforms and analytics on top of other companies' models; it has not disclosed a foundation model of its own. Its in-house assistant Red, built on Azure OpenAI Service and Azure AI Search over about 60,000 documents, reached 52,000 users by July 2025 and more than 85,000 employees by October 2025. The group is consolidating AI tools onto a single platform, UBS Claves, which offers models 'from lightweight to frontier, with automated routing that optimizes performance and token costs'. In the US, the Smart Technologies and Advanced Analytics Team (STAAT) runs the insights engine behind 'over 20 million' AI-driven insights delivered to advisors in 2025; a UBS spokesperson described it in June 2026 as 'the in-house AI tool' used by 90% of US advisory teams. Other UBS-built systems include an M&A 'co-pilot' that scans more than 300,000 companies in under half a minute (2024), RiskLab, its model development and validation platform built on Domino Data Lab's product, and analyst video avatars made with OpenAI and Synthesia models (2025). By July 2026 UBS had 560 live AI use cases and said all employees could create their own AI agents.[1][2][3][4][5][6][7][8][9][10][11]
Data as of Oct 22, 2026 · UBS profile updated Oct 13, 2026 · 24 sources cited on this page · 140 banks profiled on this site · Markdown
Which AI systems has UBS built itself, and on what models?
Systems UBS describes as created, built or developed in-house. Group systems unless marked US. 'Built on' names the outside model or platform where UBS, its vendor or the press has said so.
| Platform or model | What it does | Built on | Since | Status |
|---|---|---|---|---|
| Red | UBS-created assistant that 'integrates our extensive information library' for employees; two domain-specific assistants with multi-language support for client advisors | Azure OpenAI Service and Azure AI Search, over about 60,000 digitized documents | January 2025 (reported) | 52,000 users by July 2025; more than 85,000 by October 2025[12][2][3][4] |
| UBS Claves | Single strategic AI platform that divisions extend for their own needs; automated routing across models; AI evaluation for governance, measurement and value tracking | Range of models 'from lightweight to frontier' (providers not named) | Described 2026 | Consolidation in progress[1] |
| STAAT Insights engine (US) | Combines data analytics with machine learning to flag client and prospect events and send pre-meeting briefings to US financial advisors; an 'analytics platform built by advisors, for advisors' | In-house (UBS STAAT) | Described June 2025 | Over 20 million insights in 2025; used by 90% of US advisory teams (June 2026)[13][5][6][7] |
| Advisor agents (US) | Gather meeting, account and e-mail information, send alerts such as a maturing annuity and, after the advisor decides, execute trades and transfers | Not disclosed | July 2026 | In use[14][15] |
| M&A co-pilot | Generates buy-side ideas, finds potential buyers in sell-side situations and flags possible activist targets by comparing management tone; analyses more than 300,000 companies in under half a minute | Built by UBS (model not disclosed) | In use for about a year by September 2024 | In use; valuation and benchmarking tests 'did not yield satisfactory results'[8] |
| RiskLab and the AI Common Ecosystem | Platform for AI model development and validation used by more than 1,200 data scientists; managed LLM endpoints, RAG environments and agent deployment pipelines | Built within the UBS environment on Domino Data Lab's product, on Azure | Over five years of collaboration by August 2025 | Being rolled out across the bank[9][16] |
| Analyst video avatars | Turns research notes into scripts and avatar videos that analysts review and approve; more than 36 of about 720 analysts signed up | OpenAI and Synthesia models | May 2025 | Global rollout planned[10] |
| Global Research AI Toolkit | Generative AI across research workflows and the research value chain; Celent 2026 Model Sell Side Award for GenAI Automation | Not disclosed | Award reported July 2026 | In production[11] |
| Employee-built agents | All employees can create their own AI agents; about 18,000 developers use AI across the software lifecycle | Not disclosed | July 2026 | Available to all employees[11] |
Which UBS teams build its AI?
UBS runs a hub-and-spoke model: the Chief AI Office in the Group Chief Operating Office owns strategy, governance, regulatory engagement and 'AI Capabilities enablement', while business divisions and group functions own delivery. Daniele Magazzeni, formerly JPMorgan's chief analytics officer for EMEA, became the first chief AI officer on January 1, 2026.[1][17]
In US wealth management, STAAT, the Smart Technologies and Advanced Analytics Team inside Global Wealth Management US, builds the advisor insights engine; its stated mission is 'to build an integrated ecosystem that puts the best information together'. In investment banking, Ronald Jansen ran an AI lab in the global banking division for seven years before joining JPMorgan in early 2026.[13][5][18]
Research capacity sits in the UBS-IDSIA AI Lab, which pairs 20 IDSIA researchers with 100 UBS business domain specialists, and the Oxford-UBS Centre for Applied AI. Internal AI Kettle challenges, started in 2024 by Global Markets and Group Emerging Technology, produced nearly 300 AI solutions from 800 participants in the first round, with winning approaches implemented in the FX business.[1][19]
What does UBS build itself and what does it buy?
UBS buys the models and builds the applications and the platform that routes between them. It says it works with 'leading technology companies and select startups to ensure we have choice, resilience, and access to the best available AI models and tools', and that Azure gives it 'timely and secure access to emerging AI technologies'. Red runs on Azure OpenAI Service, the analyst avatars on OpenAI and Synthesia models, and Microsoft 365 Copilot is deployed to all employees alongside Red.[1][2][10][4]
For model development UBS chose a hybrid: RiskLab was 'built within the UBS environment, based on Domino's product', and UBS took an equity stake in Domino Data Lab in August 2025. Claves is the consolidation step, giving divisions 'shared AI capabilities and a common foundation' to extend rather than separate platforms.[9][1]
What has UBS not disclosed about its own AI?
UBS has not named the models Claves routes to, the models behind the US advisor agents or the M&A co-pilot, or whether it fine-tunes any model on its own data. It has not published Red usage after October 2025, revenue or assets attributed to STAAT, or a cost for its in-house AI.[1][14][8][4]
No patent is tied to a named UBS system. Google Patents shows two granted US machine-learning patents assigned to UBS Business Solutions AG (2022 and 2023), one listing Ronald Jansen among its inventors, and none assigned to UBS Bank USA.[20][21]
What does the UBS profile record on this topic?
The rows below are the entries of the UBS AI strategy profile that bear on this page, as of Oct 13, 2026; the profile holds the full record of 28 dated moves, 14 use cases and 35 sources.
| Use case | What the record shows | Status | Source |
|---|---|---|---|
| Copilot and the in-house assistant Red | Rolled out to all employees; 300-plus live use cases across the group in October 2025, 560 by July 2026. US advisors report Copilot usage above 85%. | In production | [17][11][22] |
| STAAT proactive insights for US advisors | Machine-learning alerts on external assets, life events and network connections. | In production | [5] |
| UBS Claves platform and AI risk committee | One platform with model routing and evaluation; AI Operating & Risk Committee under a group AI policy. | Rolling out | [1] |
| RiskLab model development and validation (Domino) | An auditable AI model-development platform built on Domino Data Lab, being rolled out across the bank. | Rolling out | [9] |
| AI avatars of research analysts | Digital twins delivering AI-generated research scripts to clients; more than 36 of 720 analysts had signed up by May 2025, using OpenAI and Synthesia models. | In production | [23][10] |
| Advisor agents for alerts and execution | Agents send thousands of daily alerts from meetings, accounts and e-mail; after an advisor decides, they can trade and complete transfers. | In production | [14] |
| Branch guidelines chatbot and meeting agents | A chatbot over thousands of internal guideline articles (roughly 20,000 advisor prompts a month), a meeting-scheduling agent, and Copilot meeting summaries said to save 20 to 30 minutes per client meeting. | In production | [22] |
| Employee-built AI agents | All employees can now create their own agents; an AI Power Hour reserves one hour a week for skills building. | Rolling out | [11] |
| Microsoft Azure OpenAI Service and Azure AI Search | Platform behind the Red assistants and a queryable knowledge base of about 60,000 investment-advice and product documents; Microsoft Azure is named as a strategic AI partner. | In production | [24][1] |
| Date | Move | Detail | Source |
|---|---|---|---|
| Jul 29, 2026 | Second quarter: 560 live use cases and a US AI award | UBS reports 560 live use cases, 920-plus in development, about 18,000 developers using AI and all employees able to build agents; its US flagship platform for AI-generated insights wins Celent's 2026 Model Wealth Manager award for Data, Analytics and AI. | [11] |
| Jul 13, 2026 | Advisor agents described to Reuters | Head of AI product Richard James says advisor agents send thousands of alerts a day and, once an advisor approves, can execute trades and money transfers; UBS says AI lets advisors spend 70% of their time with clients. | [14] |
| Mar 9, 2026 | 20-F: 20 million STAAT insights for US advisors | The 2025 annual report counts over 20 million AI-driven insights delivered to US financial advisors, nine large-scale transformational AI initiatives, and AI brought inside the model-risk framework. | [6] |
| Dec 1, 2025 | US data chief describes advisor AI | Joe Cordeira says about 80% of US advisors use STAAT Insights (400 scenarios), Copilot usage is above 85%, a branch-guidelines chatbot handles roughly 20,000 prompts a month, a meeting-scheduling agent is live and advisors save an estimated 10,000 hours a month. | [22] |
| Oct 16, 2025 | First chief AI officer appointed | Daniele Magazzeni to lead the Chief AI Office from 1 January 2026; 300-plus use cases; Copilot and Red for all employees. | [17] |
| Aug 21, 2025 | Equity investment in Domino Data Lab | UBS leads an investment in the provider behind RiskLab, its AI model development and validation platform, which is to be rolled out across the bank. | [9] |
| May 15, 2025 | AI avatars of analysts sent to clients | Scripts and digital twins built with OpenAI and Synthesia, driven by client demand for research video. | [23] |
| Dec 31, 2024 | Azure OpenAI Service reaches 30,000 UBS employees | Microsoft says its Azure OpenAI Service and Azure AI Search power UBS's Red assistants and had been rolled out to 30,000 employees as of December 2024, across wealth management, personal and corporate banking and group operations. The page is undated; the figure is as of December 2024. | [24] |
Does UBS build its own AI models?
UBS builds its own AI applications and platforms but has not disclosed a foundation model of its own. Its in-house assistant Red runs on Azure OpenAI Service, its analyst avatars use OpenAI and Synthesia models, and its Claves platform routes across lightweight and frontier models from outside providers. Its own machine learning includes the STAAT insights engine for US advisors and an M&A co-pilot that screens more than 300,000 companies.
What is UBS Red?
An AI assistant UBS created to make its information library available to employees. It is built on Azure OpenAI Service and Azure AI Search over about 60,000 digitized investment-advice and product documents, includes two domain-specific assistants for client advisors, and reached more than 85,000 employees by October 2025, per UBS's third-quarter 2025 results.
What is UBS Claves?
UBS's single strategic AI platform, onto which it is consolidating its AI tools. UBS says Claves offers models from lightweight to frontier with automated routing that optimizes performance and token costs, provides shared capabilities that divisions extend, and is paired with AI evaluation for governance and value tracking. The model providers are not named.
What AI has UBS built for its US financial advisors?
The STAAT insights engine, which combines analytics with machine learning to flag client events and prepare meeting briefings and delivered over 20 million insights in 2025, plus agents that send alerts and, after the advisor decides, execute trades and transfers (July 2026). A UBS spokesperson said in June 2026 that 90% of US advisory teams use the in-house AI platform.
What is the UBS M&A co-pilot?
An AI tool UBS built to generate buy-side ideas and identify potential buyers, able to analyse a database of more than 300,000 companies in less than half a minute and to flag possible activist targets from management tone. Brice Bolinger, head of M&A Switzerland, said in September 2024 it had been used with clients for about a year.
- Innovation and AI at UBS (undated page, retrieved 2026-10-13)
- UBS underpins AI development with data transformation initiative
- UBS reports 2Q25 net profit of USD 2.4bn and USD 4.1bn for 1H25 with integration remaining on track
- UBS reports strong results in 3Q25 with continued progress on integration
- The UBS AI Advantage (financial advisor experience)
- UBS Group AG Annual Report 2025 on Form 20-F
- The mass affluent are losing their allure for wealth managers navigating AI
- UBS Has an AI Tool That Can Scan 300,000 Firms in 20 Seconds
- UBS Invests in Domino Data Lab, Deepening Strategic Partnership to Drive Next Wave of Enterprise AI Innovation
- European bank UBS is seeing strong demand for AI-generated video avatars of its analysts
- UBS reports 2Q26 net profit of USD 2.8bn and USD 5.8bn for 1H26 (Business Wire)
- Artificial intelligence at UBS (retrieved 2026-10-23)
- Grow your practice with data and smart technology (STAAT)
- Wall Street banks ramp up digital assistants in bid to win productivity race
- Wall Street banks promoting AI agents from research aids into digital coworkers
- UBS unlocks advanced AI techniques with PostgreSQL on Azure (UBS RiskLab authors)
- UBS appoints Chief AI Officer to drive its AI strategy
- JPMorgan hired the head of UBS' AI lab to assess employee performance
- Our AI Kettle is ready to boil
- US11775887B2: Methods and systems for processing data having varied temporal characteristics to generate predictions related to management arrangements using random forest classifiers
- US11429879B2: Methods and systems for identifying dynamic thematic relationships as a function of time
- Behind UBS' AI strategy to win wallet share, client leads
- Executives Using AI Avatars Face a Patchwork of Regulations
- UBS and Microsoft unite: co-creating the future of banking with Azure AI (undated; figures as of December 2024)
When UBS 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