Built in Singapore to solve the agent infrastructure problem

The agent infrastructure gap

In 2022, AI agents were starting to work. GPT-3 function calling opened the door; early LangChain experiments showed what was possible. But every team building something real ran into the same wall: the infrastructure didn't exist.

Memory was duct-taped together with custom code. Orchestration was a pile of if/else. Observability meant printing to stdout and hoping. Deploying an agent to production meant re-inventing the same tooling every team was building in parallel.

Diaflow is the platform we wished existed. Opinionated enough to give your agents a production-ready foundation. Open enough to work with any LLM, any vector store, any tool you need.

The team

Jonathan Viet Pham

Founder & CEO

Previously built agentic pipelines at a Singapore fintech. Spent 3 years debugging orchestration code before deciding the problem needed an infrastructure solution.

Priya Krishnaswamy

Head of Engineering

Distributed systems engineer. Designed the orchestration execution engine and memory isolation architecture. Formerly led backend infrastructure at a Series B SaaS company.

Minh Duc Tran

Lead Developer Relations

Wrote his first LLM agent in 2021, has been writing about agent patterns since. Runs the Diaflow developer community and creates the docs and guides that ship with every release.

Li Ying Chen

Platform Engineer

Works on the integrations layer and the Tool SDK. Built the first version of the custom tool decorator system. Background in API design and developer tooling.

Company facts