Portable by default
Agents, profiles, and workflows should move without being rebuilt around a single vendor.
Alex Merced on agentic AI
I build open tools, specifications, and ideas for agentic systems that people can understand, govern, and move.
THE THESIS
“The future of agentic AI is not one model, one agent, or one platform. It is replaceable components connected by open contracts.”
SELECTED OPEN WORK
Projects spanning brokerage, execution, memory, governance, and interoperability.
A provider-neutral broker for routing work across agents and model-powered tools.
View repository ↗An agent harness built around explicit authority, policy, evidence, and durable records.
View repository ↗A practical Python agent framework for composing providers, tools, memory, and workflows.
View repository ↗A graph-shaped memory layer for representing relationships, context, and retrieval paths.
View repository ↗OPEN CONTRACTS
Open software is strongest when its components share a durable language. These specifications focus on graph-shaped work and portable agent identity.
DESIGN PRINCIPLES
Agents, profiles, and workflows should move without being rebuilt around a single vendor.
An agent should know what it may do, what requires approval, and where its responsibility ends.
Memory, context, and execution history should be understandable—not trapped behind a magic interface.
Useful autonomy comes from traceable decisions, observable work, and verifiable outcomes.
ABOUT ALEX
Alex Merced works across data infrastructure, developer education, and agentic AI. His focus is making complex systems legible—and giving builders open foundations they can adapt, audit, and own.