Hi everyone! I’m Sherry, the founder of Custodian Labs, and I’m excited to finally share what my team has been building. Everyone talks about creating AI agents, but getting it hosted, connected to company data, reliable in production, and safe to use with sensitive information is a completely different challenge. We built Custodian Labs to collapse that entire process into a simple SDK. With Custodian, developers can: - Build and deploy an AI agent in around five lines of code - Add RAG without configuring embeddings or a vector database - Switch between a lot of models! (Like, 100+) - Create specialised multi-agent teams with automatic routing - Detect, mask, redact, or transform PII before it reaches an AI model The last part became especially important to us. Most privacy tools simply remove sensitive information, which can also remove the context an AI system needs. Our Guardian Layer can replace PII with realistic synthetic equivalents, allowing the model to retain useful context without receiving the original sensitive data. Our goal is to let developers focus on what their agents should do! Not databases, infrastructure, hosting, model integrations, or privacy plumbing. We’re still early, so I’d genuinely love your feedback: What would you need to see before trusting a platform like this in production? Which integrations or capabilities should we prioritise next? Thanks for checking out Custodian Labs!

Custodian Labs

Deploy AI Agents in 5 lines of code

Deploy AI Agents in 5 lines of code