From 5,500+ agents to 642 models, from multi-agent orchestration to durable execution, from full-chain observability to agent protection — Metaprise is the most complete Agent OS for enterprise AI. All tools free. Pay only per Mission.
PLATFORM CAPABILITIES
Nine integrated modules that work together as one unified Agent OS. From agent marketplace to full-chain observability, from durable execution to agent protection — each module is production-grade, enterprise-hardened, and available across all three deployment modes.
Every engine, every agent, every policy — managed from one unified Dashboard. You define the rules. The Dashboard translates them into execution constraints that apply to all agents, in real time, without code changes.
5,500+ open-source Agents, MCP plugins, and enterprise SaaS integrations in one unified marketplace. One-click install to Cloud, Hybrid, or Enterprise. Every agent undergoes L1–L4 risk assessment and is wrapped by OrgKernel's ExecutionToken boundaries and AuditChain logging. 1,200+ official MCP servers and 3,200+ ClawHub skills.
BROWSE AGENTS642 models from 40+ providers through one unified API — LLM 380, Vision 98, Embedding 62, Reranker 18, Image Gen 45, Audio 22, Reasoning 35. 182 models available via Serverless inference on pay-per-token billing. Side-by-side comparison of up to 4 models with built-in cost estimation. Version Lock freezes model behavior for reproducible agent output.
EXPLORE MODELSTruth · Compliance · Execution · Audit · Business State Machine · Knowledge Graph. Six production-grade engines form the execution layer of the Agent OS. Each engine is configurable at runtime through the Dashboard — no code changes required.
EXPLORE ENGINES9 built-in multi-agent architectures: ReAct, Plan-and-Execute, Tree-of-Thought, Reflexion, Graph/DAG, State Machine, Multi-Agent Systems, Tool-Router, and Memory-Driven. Time-Travel Debugging rewinds to any prior state. Cross-Agent RemoteGraph unifies local and remote agent calls. Nexus RPC enables cross-namespace, multi-region routing.
VIEW ARCHITECTURESAmerica's third open-source LLM (after Llama and DBRX), built on a hybrid Llama architecture and trained on 8+ years of financial, medical, and legal compliance data. Version Lock freezes agent behavior for predictable, reproducible output. Every inference pass is logged through AuditChain for full auditability.
EXPLORE LLMThe core governance runtime — OrgKernel trust layer (Ed25519 identity, ExecutionToken, SHA-256 AuditChain), Mission Layer with 8-state execution machine, Policy Engine with deterministic PERMIT/DENY/ESCALATE, and L1–L4 Authority Graph. Durable Execution with crash recovery, Saga Compensation, Checkpoint Persistence, Retry Policy Engine, Namespace Isolation, multi-region replication (99.99% SLA), Workflow Versioning + Canary releases, SSO/SAML, and SCIM user sync.
EXPLORE AURAPython, TypeScript, Go, and Java SDKs with full Mission/Agent/Audit/Registry API. CLI tools, MissionRegistry distribution protocol, and Mission Store marketplace with 30% revenue share + Token Royalty Protocol. 600+ enterprise connectors (Stripe, PayPal, SAP, Oracle, Salesforce, HIPAA, HL7). One-click Deploy CLI and No-Code Agent Builder for non-technical teams.
START BUILDINGBuilt-in full-chain tracing visualizes every agent decision, tool call, and model response. Online and offline evaluation with LLM-as-judge scoring. Insights Agent automatically clusters failure patterns and surfaces emerging issues. AI Debug Assistant, Annotation Queue for human review, Prompt Playground, A/B experiments, cost/latency/error dashboards, alert rules (PagerDuty/Slack/Webhook), and OpenTelemetry export.
EXPLORE OBSERVABILITYBuilt-in agent protection layer that safeguards context health during long-running tasks. Context summarization prevents context window pollution, loop detection middleware intervenes after N iterations, pre-completion checklist validates against original task spec, virtual filesystem backend offloads tool results to storage, and lifecycle hooks provide pluggable before/after middleware around every agent action.
EXPLORE HARNESSDEPLOYMENT MODES
Every deployment mode runs the same Agent OS runtime. Your agents, orchestration logic, and business rules are portable. Migrating from Cloud to Enterprise is a configuration change, not a rebuild.
Speed
Metaprise hosts everything. You sign up, install agents from the store, and start running enterprise workflows on the same day. No ops team, no GPU setup, no cold starts.
Balance
Metaprise manages the control plane. Your data plane runs inside your own network perimeter. Sensitive data never crosses your boundary — even at rest.
Sovereignty
You own the entire stack — code, models, data, runtime. Zero data egress. Optional air-gap mode for environments with no external network connectivity whatsoever.
PLATFORM ARCHITECTURE
One codebase, three deployment boundaries. The control plane manages orchestration and policy. The data plane executes agents and processes your data. Only the boundary moves — the runtime stays identical across all three modes.