Starling Bank Debuts 'Talk to Transact' Voice Payments via Agentic AI

Starling Bank is introducing voice-activated payment capabilities to its mobile app, allowing retail and joint account holders to initiate transfers using natural language commands. This shift toward "agentic commerce" aims to remove the administrative friction of digital banking by replacing manual data entry with an AI-driven interface that handles the setup of transactions.

What was announced

The new "talk to transact" feature, powered by the Starling Intelligence platform, allows customers to set up payments by speaking directly to their phone. By using prompts such as “Pay the plumber £200” or “Send Mum £20 for lunch,” the in-app Starling Assistant automatically configures the payment details. The customer then reviews the information and completes the transaction with a single screen tap. This capability is scheduled to be rolled out to all retail customers by the end of 2024.

A similar feature is also in development for small business customers. This iteration will enable users to instruct the Starling Assistant to “Pay this invoice” while uploading an image of a bill, allowing the AI to extract relevant payment data and prepare the settlement. These updates accompany a broader suite of "Smart Tools"—pre-configured AI prompts that Starling began releasing in August. To date, the bank has launched 14 such tools, including a holiday spending analyser and a 50/30/20 budget planner. Starling has committed to releasing one new smart tool every week until the end of the year, with future functionality expected to allow customers to build their own bespoke tools within the app.

"Agentic payments are following mobile wallets as the next evolution of payments here in the UK. Gone is the need for multiple payment screens and laboriously inputting payment details. Now, payments and transfers can happen with a simple natural language conversation,"

Harriet Rees, Group CIO at Starling.

The companies involved

Starling Bank is a prominent UK challenger bank that has transitioned from a mobile-only startup into a major player in the retail and SME banking sectors. Unlike many fintechs that operate as electronic money institutions, Starling holds a full UK banking license and manages its own proprietary technology stack. This vertical integration allows the bank to deploy complex AI features, such as the Starling Intelligence platform, directly into its core banking environment.

The bank’s technical strategy is led by Harriet Rees, Group Chief Information Officer. Rees holds a significant role in the wider UK financial landscape as a Government AI Champion for Financial Services. In this capacity, she co-authored the Financial Services AI Adoption plan, a strategic document focused on the modernisation of UK payment services through the integration of agentic commerce. Starling’s current focus on agentic AI—where software agents perform tasks on behalf of a user rather than simply providing information—is a central component of its effort to differentiate itself from both traditional high-street lenders and other digital-first competitors.

What FF News has reported before

FF News has extensively covered Starling’s evolution, with 119 stories tracking its growth and product expansion. In August 2025, the bank significantly enhanced its utility for entrepreneurs when Starling Agrees to Buy Ember to Add Digital Tax and Accounting Tools for Small Business Customers. This acquisition was followed by a major brand refresh, as Starling Reveals New-Look Logo, App and Cards as Bank Launches Brand Mission to Help Britons Become ‘Good With Money’. More recently, the bank has sought to increase its public profile through sports marketing, as Starling and Arsenal Launch New Partnership With Financial Literacy Campaign.

What this means

This move into agentic payments places immediate pressure on the wider UK banking sector to move beyond basic chatbots. While many institutions use AI for back-end fraud detection or surface-level customer service, Starling is attempting to make AI the primary interface for the transaction itself. The success of "talk to transact" will depend heavily on the reliability of natural language processing in high-stakes financial contexts; any persistent errors in voice recognition could quickly erode user trust. However, if successful, it sets a new industry benchmark for "invisible banking," where the technical complexity of moving money is entirely abstracted away from the user experience.