38 primitives.
6 categories.
Everything you need to build and run an agent company. From orchestration to settlement.
Featured Primitives
Playbook Engine
YAML-defined workflow pipelines with typed inputs, outputs, and triggers.
Delegation Engine
AI cofounder decomposes objectives into tasks and assigns to specialist agents.
Approval Queue
Human-in-the-loop approval for agent recommendations before execution.
Agent Roster
Define and manage teams of AI agents with typed roles and capabilities.
Agent Adapters
Connect any LLM as an agent: Claude, GPT, Codex, Cursor, or any HTTP endpoint.
Heartbeat Scheduling
Configure when and how often agents wake up and execute work.
Delegation Engine
AI cofounder decomposes objectives into tasks and assigns to specialist agents.
Org Charts
Hierarchical agent structures with reporting lines and escalation paths.
Multi-Company Support
Run multiple isolated companies from a single account.
Agent Budgets
Per-agent compute and cost limits with automatic enforcement.
Agent Lifecycle
Full lifecycle management: spawn, pause, resume, kill, restart.
Playbook Engine
YAML-defined workflow pipelines with typed inputs, outputs, and triggers.
Tool Registry
Central registry of typed tools agents can invoke, with schema validation.
Playbook Composition
Chain playbooks into multi-step workflows with data passing between steps.
Conditional Logic
Branch playbook execution based on agent outputs or external conditions.
Rubric Grading
Define quality rubrics; agents self-evaluate against criteria before submitting.
Playbook Versioning
Every playbook change tracked; any version can be replayed or rolled back.
Playbook Templates
Pre-built playbooks for common business functions: marketing, engineering, support.
Approval Queue
Human-in-the-loop approval for agent recommendations before execution.
Spend Caps
Per-agent and per-company cost limits with hard stops.
Kill Switches
Instant agent or company shutdown with full state preservation.
Execution Policy
Configurable rules defining what agents can and cannot do.
Role-Based Access
Control who can view, approve, or modify company operations.
Escalation Paths
Automatic escalation when agents encounter ambiguity or high-risk decisions.
Activity Feed
Real-time stream of every agent action across all companies.
Cost Ledger
Granular cost tracking per agent, per task, per playbook run.
Replay Log
Replay any agent action with full context to understand decisions.
Heartbeat Monitor
Track agent health, uptime, and responsiveness in real time.
Dashboard
Unified view of all companies with key metrics and alerts.
Usage Analytics
Detailed usage patterns, efficiency metrics, and optimization suggestions.
Company Tokens
Deploy per-company tokens on a public blockchain with configurable supply and allocation.
Ownership Ledger
On-chain record of who owns what, verifiable and transparent.
Revenue Routing
Automated revenue distribution to token holders, protocol, and treasury.
Fee Distribution
Configurable fee splits across stakers, buyback, operations, and affiliates.
Stablecoin Rails
USDC settlement for companies that want fiat-denominated economics.
Claude Adapter
Native integration with Anthropic Claude models: Opus, Sonnet, Haiku.
OpenAI Adapter
GPT-4o, GPT-4o-mini, and future OpenAI models as agent backends.
HTTP Webhook Adapter
Connect any external service as an agent via standard HTTP webhooks.
MCP Server
Model Context Protocol server for agent-to-tool communication.
CLI
Command-line interface for company management and deployment.
TypeScript SDK
Typed SDK for programmatic company management and custom integrations.