Skip to main content

What is Piper?

Piper is an enterprise-grade agent platform for building and deploying AI-powered task automation. Unlike chatbots that focus on instant responses, Piper agents handle complex, long-running processes that require deep work over minutes or hours.

Key Capabilities

Long-Running Processes

Handle tasks that take minutes or hours, not seconds. Agents can work asynchronously while you focus on other things.

Internal System Access

Connect agents to your internal APIs, databases, and tools. Extensive integration capabilities for enterprise environments.

Async Delegation

Agents can delegate work to other agents, creating powerful automation networks that mirror how teams actually work.

Enterprise Security

Deploy in your VPC, on-premises, or air-gapped environments. Bring your own LLM keys for complete data control.

Architecture

Piper uses an agent-graph model where agents are autonomous entities that:
  • Have relationships with other agents (can delegate and collaborate)
  • Have access to specific tools and knowledge bases
  • Emit events that channels decide how to surface
  • Persist state and can resume from any point
This architecture is inspired by how humans work in organizations: some people have each other’s contact information, some have access to certain tools, and some have specific domain knowledge.

Deployment Options

ModelLLM KeysUse Case
VPCCustomer brings their ownEnterprise with strict data requirements
SaaSWe provide (usage-based billing)Simpler deployment, managed infrastructure
On-premCustomer brings their ownAir-gapped environments

Core Primitives

PrimitiveDescription
AgentsAutonomous entities with prompts, tools, knowledge, and relationships
ToolsCode executed in a sandbox at runtime for system integration
ChannelsMessage sources (Slack, Teams, email, webhooks)
KnowledgeDocument collections with vector search
GuardrailsConstraints preventing risky or unauthorized actions

Next Steps