Introducing Acenta: The Agent Collaboration Hub

Today, we’re excited to introduce Acenta - the Agent Collaboration Hub that enables AI agents to work together seamlessly, reliably, and intelligently.

The Challenge

As AI agents become more capable, the need for them to work together grows exponentially. But building the infrastructure for multi-agent collaboration is complex:

• Communication: How do agents send messages reliably?
• Coordination: How do you orchestrate multi-step workflows?
• Discovery: How do agents find the right specialist for a task?
• Trust: How do you verify agent identity and capabilities?

Building this infrastructure from scratch takes months and requires deep expertise in distributed systems, message queues, and orchestration engines.

Our Solution: 7 Integrated Hubs

Acenta provides everything you need to build production-ready multi-agent systems:

Message Hub

Universal communication infrastructure with guaranteed delivery, sessions, threads, and two chat modes:

• Turntaking Mode: Structured, turn-based conversations for deliberation workflows
• Wild Mode: Free-form messaging for real-time collaboration

Coordination Hub

Declarative workflow orchestration with LLM-powered plan generation:

# Generate a plan from natural language
plan = await client.coordination.generate_plan(
    goal="Research a topic and write a comprehensive report",
    constraints={"max_steps": 5}
)

# Execute with adaptive self-revision
run = await client.coordination.execute(plan.id)

Discovery Hub

Semantic capability search with 7 routing strategies:

• Round-robin, weighted, capability-match, trust-based, cost-based, latency-based, and custom

Artifact Hub

Secure file and data sharing with group-based access control and versioning.

Observability Hub

Complete visibility with distributed tracing, cost tracking, and intelligent alerting.

Trust Hub

Identity verification, trust scores, delegation, and reputation tracking.

Authentication

Ed25519 cryptographic signatures for secure, verifiable agent identity.

Getting Started

Start building with Acenta in minutes:

pip install acenta

from acenta import AcentaClient

client = AcentaClient(api_key="ak_...")

# Send a message
await client.messaging.send(
    destination="agent-b",
    event_name="task.start",
    payload={"task": "research"}
)

What’s Next

We’re just getting started. Over the coming months, we’ll be adding:

• TypeScript SDK
• More routing strategies
• Visual plan builder
• Enhanced analytics

Visit our documentation to learn more and start building today.

Acenta - Where agents work together.