BlackRock, the world’s largest asset manager with $14 trillion in assets under management, has launched RockAI, a no-code platform that lets in-house developers build and deploy custom AI agents without writing code. The platform rolled out to BlackRock’s 5,000 developers last week, according to AI Street, citing reporting from the Wall Street Journal.
How RockAI Works
RockAI serves as the centralized interface for all AI agents built inside BlackRock. Its natural language interface lets users select AI models, provide context, and connect to internal databases to spin up task-specific agents, according to the WSJ. Pavan Pemmaraju, a senior lead in software engineering on the Aladdin Product Engineering team, told the WSJ that safety and security guardrails are built in, so users can create new agents in minutes and have them ready to scale across the business.
Nish Ajitsaria, BlackRock’s head of Aladdin Product Engineering, told the WSJ that AI will become the default mode for most processes, including research, coding, and operations. Humans will shift into smaller, cross-functional teams that oversee the work rather than do it themselves.
From Developers to Citizen Developers
The initial rollout targets BlackRock’s engineering staff, but the firm plans to extend RockAI to non-technical employees. The goal: let those in nontechnical roles, dubbed “citizen developers,” build agents that can replace chunks of their own manual work and share them with colleagues, according to AI Street.
The use cases span investment research, data analysis, and risk modeling. One example cited by the WSJ: an agent designed to research the top two real-estate investment trusts in the U.S., built by selecting a model, entering the relevant context, and connecting to the necessary databases.
The Enterprise Agent Infrastructure Pattern
BlackRock’s approach reflects a pattern emerging across large financial institutions. Rather than buying agent capabilities from vendors or running one-off pilot projects, firms are building internal platforms that treat agent development as a standard organizational capability. The no-code interface addresses a practical constraint: most enterprises lack specialized agentic AI engineering teams, so they need to let domain experts build their own tools.
The distinction matters. A centralized agent platform with built-in governance is a different proposition than individual teams experimenting with ChatGPT or standalone coding assistants. It implies a bet that agent building will become as routine as spreadsheet creation, not a specialty reserved for AI engineers.