JPMorgan Chase
LLM Suite internal generative AI platform
JPMorgan's in-house AI platform reached 200,000 employees in eight months without being mandated
Problem
JPMorgan Chase wanted its staff to use large language models without sending sensitive data to consumer AI services, which employees might otherwise turn to on their own.
Approach
In summer 2024 the bank released LLM Suite, its own platform giving eligible employees secure access to third-party large language models for drafting, summarising and idea generation. Employees chose whether to use it. The bank updates it about every eight weeks, adding connections to its data and software, and has started building agents for multistep tasks.
Results
Takeaway
Staff adopted the tool voluntarily and quickly, while the bank's expectation of fewer operations staff made job redesign a management issue. The bank has not published a measured productivity result.
Categories
- Size
- Large enterprise
- Industry
- Finance and insurance
- Business function
- Internal work and knowledge managementStrategy and organization
- Innovation type
- Organization and ways of working
- AI technology
- Generative AI
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
JPMorgan reports that LLM Suite won American Banker's 2025 Innovation of the Year award. It says the platform was released to eligible employees in summer 2024 and grew to 200,000 onboarded users within eight months. Employees use it for drafting and idea generation, and the bank plans to connect more internal data and build agents.
AI News reports that JPMorgan built its own system for security and control. It says LLM Suite reached about 200,000 users eight months after launch and quotes chief analytics officer Derek Waldron on how the opt-in approach drove "viral adoption". It cites his remark to McKinsey that a little under half of employees use generative AI tools every day, and reports that the bank expects operations staff to fall by at least 10%.
TipRanks, citing Derek Waldron's interview with CNBC, reports that LLM Suite is updated every eight weeks with data from the bank's databases and software, and that JPMorgan aims to become a fully AI-powered enterprise. Waldron says the change will take years because of a gap between what AI can do in theory and in practice. A demo showed the platform building an investment banking presentation in about 30 seconds.
Full details are in the original sources.