🔍 Read the full analysis: UBS’s AI Agents At Work: What’s Known And What’s Unclear on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
UBS reported on October 28 that employees were running more than 12,700 personal AI agents, with 608 AI use cases in production and more than 1,090 in development. The bank has not explained what those agents can access or do, how they are monitored, or how many are used in client-facing work.
UBS said on October 28 that its employees were running more than 12,700 AI personal agents to support daily work, putting a concrete deployment figure into the Swiss bank’s third-quarter results. The disclosure also listed 608 AI use cases in production and more than 1,090 in development, but did not specify what the personal agents can do, what information they can access or how the bank oversees them.
In its third-quarter results release, UBS reported nine large-scale AI initiatives, 608 AI use cases in production and more than 1,090 further applications in development. The bank said its live use cases were up 81% from a year earlier. It also said more than 18,000 software developers were using AI tools in software development, describing the productivity gains as measurable but providing no amount or method for measuring them.
UBS said employees had generated more than 30 million insights through STAAT so far in 2026. The tool supplies suggestions to the bank’s U.S. financial adviser teams. The results release did not say how many of those insights were acted on or what effect they had. Reuters reported the 12,700-agent figure on October 28, and PYMNTS repeated it the following day; neither account added operational detail beyond the bank’s disclosures.
The figures show a substantial internal AI program, but they describe different things: a count of agents, production use cases, projects in development and employees using developer tools. UBS did not explain whether these categories overlap, how it counts an agent, or whether the 12,700 figure means agents created, currently active, or used over a particular period.
Infographic 01 · What UBS reported
12,700 personal AI agents, inside one bank
The AI figures from UBS’s third-quarter results, and how fast they grew in three months.
From permission to 12,700 agents
UBS 2Q26 release (29 Jul 2026) vs. 3Q26 release (28 Oct 2026)
What the agent number is not
- Not 12,700 people. One employee can run several agents; UBS does not say how many people use them.
- Not a definition. UBS does not say what an agent can access or do on its own.
- Not a result. “Measurable productivity gains”, but no hours or costs published.
- Scale, for context: about 13 agents per 100 internal full-time staff (12,700 ÷ 97,593, own calculation).
“Our employees are running over 12,700 AI personal agents to support them in their daily activities.”UBS, third-quarter 2026 results release, 28 October 2026
Also in the release: 9 large-scale AI initiatives and more than 30 million insights from the STAAT tool for US financial advisor teams so far this year. UBS placed 6th of 50 banks in the 2026 Evident AI Index.
Sources: UBS 3Q26 release, 28 Oct 2026 · UBS 2Q26 release, 29 Jul 2026. Reported identically by Reuters, 28 Oct 2026 and PYMNTS (29 Oct 2026).
Powered by Thorsten Meyer AI
What 12,700 Agents Mean for Oversight
The disclosure matters because it shifts the discussion from whether a large bank is experimenting with AI to how many tools its employees are already using. In a regulated institution, the practical questions concern data access, permissions and human review, as well as whether an agent only prepares information or can take an action.
Those distinctions affect both customers and staff. An assistant that drafts a summary for an employee presents a different risk from a system that can send a message, update a record or initiate a transaction. UBS has not said which capabilities its personal agents have. The results therefore establish the scale the bank reports, not the degree of automation or the effect on customer outcomes.
For other businesses, UBS’s disclosure offers a measure to compare against as they set rules for employee-built AI. The count alone does not show whether the bank has effective controls or whether the tools deliver savings. It does make governance a concrete operational issue: firms need to know what is being used, by whom and under what limits.
Products Worth Considering
LG UltraGear 27-Inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDMI 2.1
LG UltraGear 32GX850A-B 32-Inch 4K UHD Glossy OLED Gaming Monitor
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From Employee Access to Reported Use
UBS had signaled wider employee access before reporting the 12,700 figure. In its July 29 second-quarter release, the bank said all employees had the ability to create their own AI agents. UBS said the aim was to deploy AI safely and at scale while allowing staff to spend more time on judgment, collaboration and client work. Access to a tool, however, does not establish how many employees used it or what tasks it performed.
The bank’s technology information also describes Red, an in-house assistant made available to around 100,000 employees, and an AI-powered HR assistant in Microsoft Teams for tasks such as checking policies and booking holidays. UBS has not said whether these products are included in the personal-agent count or how they relate to the 608 production use cases.
UBS has drawn a clearer boundary around one higher-risk category. At an event at ETH Zurich on October 27, CEO Sergio Ermotti said the bank does not allow outside AI models to enter the bank and act for clients, such as managing accounts or moving deposits, according to Reuters. That statement concerns external models acting on clients’ behalf; it does not define the capabilities or permissions of UBS’s employee-run agents.
“Employees “are running over 12,700 AI personal agents to support them in their daily activities.””
— UBS, third-quarter results release, October 28, 2026
What UBS Has Not Explained
The results release does not define a “personal agent” or say which models and software power the tools. UBS has not disclosed what internal systems or customer information they can access, whether they can take actions without an employee’s approval, or what review and monitoring procedures apply.
It is also unclear how the 12,700 figure is measured: the bank gives no period, active-use rate, count of employees using agents or breakdown by task. UBS does not say how many of its 608 production use cases are customer-facing. Its statement about excluding outside models from acting for clients does not answer whether internal AI tools support work on client accounts or how those tools are controlled.
The release also does not connect AI deployment to staffing changes. UBS’s internal headcount fell by about 1,500 full-time equivalents in the quarter to 97,593. UBS and Reuters attributed the decline to the Credit Suisse integration; UBS did not say it was caused by AI. The available information does not establish a link between the agent count and employment changes.
Swiss financial regulator FINMA has said banks should maintain a complete inventory of AI applications with risk classifications, and that keeping track has become harder as applications become accessible to more people. The source material does not establish whether FINMA has made any finding about UBS’s agent program.
Further Disclosure Will Set the Picture
The next useful detail would be an explanation from UBS of what it counts as an agent and how it controls access to data and actions. A breakdown of the tools’ permitted tasks, approval requirements, employee usage and client-facing applications would help show whether the reported deployment is mainly administrative support or extends into more consequential work.
UBS’s results release does not announce a separate timetable for publishing those details. The bank’s future reporting on AI applications, productivity measures and risk controls may clarify the scale and effects of the program. Until then, the confirmed picture is limited: UBS reports more than 12,700 employee-run agents, but their capabilities and oversight remain unspecified in the disclosure.
Key Questions
What did UBS report about AI agents?
UBS said employees were running more than 12,700 AI personal agents to support daily activities, in its October 28, 2026, third-quarter results release.
What can the UBS agents do?
The bank has not defined the agents’ capabilities. Its release does not say what systems or data they can access, or whether they can take actions on their own. Examples such as drafting summaries or sorting work are possibilities, not confirmed descriptions of UBS’s tools.
Can outside AI models act for UBS clients?
CEO Sergio Ermotti said on October 27 that UBS was not allowing outside models to enter the bank and act for clients, such as managing accounts or moving deposits. That statement does not describe the permissions of UBS’s internal employee agents.
Did UBS link AI agents to job cuts?
No. UBS’s internal headcount fell by about 1,500 full-time equivalents in the quarter, to 97,593, and UBS and Reuters tied the decline to the Credit Suisse integration. UBS did not attribute the change to AI.
Source: ThorstenMeyerAI.com
