Lloyds launches Envoy platform for building and sharing AI agents
Key Points
- Lloyds launches Envoy: A secure internal Google Cloud platform for teams to easily build, train, share and monitor governed AI agents with templates and an internal marketplace.
- Key features: Built-in safety controls, human oversight, audit trails, memory for seamless customer journeys, and reuse across the bank to cut duplication.
- Business goals: Boost colleague productivity, improve customer experiences, enable new models — all while prioritising trust, safety and accountability.
- Peers in action: NatWest offers AI Digital Enabler and agentic engineering; Barclays rolls out Colleague AI Agent via Copilot; HSBC uses Mistral-powered productivity tools.
- Industry context: Part of a wider UK banking push into responsible agentic AI, with Lloyds’ dedicated agent-building platform currently standing out.
Lloyds Banking Group has launched Envoy, an internal platform which provides a secure and governed way to build and run AI agents across the organisation.
The platform aims to help the bank scale “agentic AI” responsibly while improving both customer experiences and internal productivity, it said.
Envoy provides ready-made templates so teams don’t have to start from scratch. It connects to Lloyds’ existing Large Language Model platform, enforcing safety rules, risk checks and human oversight before agents go live.
Once deployed, the system offers continuous monitoring, full visibility and a complete audit trail of every agent’s activity, the group said.
Straightforward with memory tools
Built on Google Cloud, Envoy is engineered for scale. Teams can publish finished agents to an internal Agent Marketplace, allowing other parts of the business to discover, reuse and build on proven solutions rather than duplicating effort.
The platform lets agents retain relevant details across conversations while respecting strict data privacy and retention rules. This should reduce the need for customers to repeat information when following up on enquiries, the bank said.
“Envoy helps our employees become more productive, improve customer journeys, and launch potentially disruptive business models,” said Ron van Kemenade, Chief Operating Officer at Lloyds Banking Group.
The launch comes as Lloyds looks to expand the value it generates from AI. The group previously reported around £50 million in value from generative AI in 2025 and expects more than £100 million this year.
How other UK banks are playing the AI agent game
Lloyds isn’t the only high-street name going hard on governed AI – but its dedicated agent-building platform with a built-in marketplace sets it apart for now.
NatWest has rolled out its own internal ‘AI Digital Enabler’ to 99% of workers, an in-house AI tool for secure experimentation with summarising documents, brainstorming and problem-solving.
It’s also pushing agentic AI hard after trials of “agentic-led engineering” delivered 10x productivity gains in financial crime teams, and it’s testing a customer-facing agentic assistant inside its Cora digital helper.
Barclays is deploying a custom ‘Colleague AI Agent’ powered by Microsoft 365 Copilot to 100,000 staff globally. The tool handles everything from travel booking and HR queries to compliance checks, all wrapped in the bank’s own governance layer.
Barclays talks openly about moving into “emerging agentic AI patterns” for both internal workflows and customer solutions, backed by a GenAI Centre of Excellence.
HSBC gives all colleagues access to an LLM-based productivity platform and has deployed generative AI assistants in customer service and corporate banking. It uses AI Review Councils for oversight but hasn’t announced a full agent-building marketplace yet.
Smaller players are also moving fast on the customer side. Starling Bank claims to have developed the UK’s first agentic AI financial assistant for personal money management, while Santander has already run Europe’s first live end-to-end payment executed by an AI agent in partnership with Mastercard.
