Show HN: Multi-agent platform to create and manage agents that do work for you

i think ai agents should be declarative, not imperative

building agents with DAGs, Graphs, and Chains means you're forcing imperative guardrails instead of allowing the agent to reason about which step to take in what order. i believe these will soon be considered tech debt as models like o1 are widely adopted

building a team of declarative reasoning agents means providing instructions and tools, and allowing them to communicate with each other to accomplish work

i want to make building these multi-agent teams and then deploying and monitoring them in production easy for anyone (not just engineers), so i'm building aster agents

bring your own llm api keys and connect integrations to unlock tool calls

for custom tool calls, you can bring your own API to dispatch them

otherwise, you don't need code