Threesixty
Production platform

Governed AI operations your leadership can trust

Fleet visibility, multi-agent Gateway architecture (up to 20+ agents per machine), WASM skill sandboxing where enabled, safety guardrails with human-in-the-loop, and controlled rollouts—so autonomous agents stay operational, secure, and within budget after pilots become production.

Book a live demo

One place to see health, act fast, and keep agents running.

Fleet operations

  • Fleet-wide visibility into health, incidents, costs, and early warning signals before the next failure.
  • Predictive forecasting on CPU, RAM, and disk from metrics history so capacity issues surface before agents fail.
  • Operator actions when things break: restart, rollback, restore from backup, and escalate with context.
  • Configurable escalation policies and on-call rotation so the right owner is paged with full incident context.
  • Secure connectivity so ops can reach every agent without agents reaching each other.
  • Optional dedicated private mesh per customer so your agents never share a network with other tenants.
  • Nightly drift scoring with behavioral, semantic, and environment dimensions—history charts and fleet heatmaps before customers feel degradation.
  • Swarm drift detection flags correlated degradation across agents on a host or customer cohort—not just isolated outliers.
Technical details for security teams
  • Command Center: fleet dashboard, per-machine and per-agent actions, Incidents (escalation, ClawGuard, resilience), releases, skills, billing, and immutable audit log.
  • Tailscale hub-and-spoke ACLs; one Gateway on :8000 routes multiple agent runtimes per machine; dedicated customer mesh available for sovereign isolation.
  • Scheduled per-agent backups with offsite copy; restore to last known good or rollback code; aggregated health per machine.
  • Composite drift scores, 7–30 day history charts, continuity snapshot timeline with manual realign, and a fleet Agent Quality heatmap.
  • Nightly fleet operations: drift scoring, known-good snapshot capture, graduated auto-realign (L1–L5) with approval gates; trajectory spans with trace IDs and cost attribution.

Up to 20+ agents and 10 claw types behind one Gateway per machine—the structural edge single-bot vendors cannot match.

Multi-agent fleet architecture

  • Run heterogeneous agents (OpenClaw, NanoClaw, and eight other claw types) on one host without fragile monolith sprawl.
  • WASM-isolated custom skills with capability tokens so executable logic never runs directly on the host filesystem.
  • Cross-agent prompt context cache on the Gateway cuts duplicate tokens and latency when agents share long system prompts.
  • Staging resilience drills produce a client-facing scorecard—proof your guardrails detect injection and alert paths before production waves.
Technical details for security teams
  • Gateway aggregates health, backups, ClawGuard middleware, and context-cache metrics to Command Center heartbeats.
  • OpenClaw WASM tier: wasmtime sandbox, manifest + capability tokens, catalog install gates for zip hardening.
  • Gateway context cache (observe + optimize modes); Command Center reports cache savings per machine.
  • Resilience scorecard in the client portal; operator resilience validation history from staging drills.

What your customers and procurement teams actually see.

Client portal

  • Branded portal per customer with live machine status, SLA tracking, and uptime they can trust.
  • Support tickets with priority tracking and satisfaction feedback after resolution—no shadow email threads.
  • Client-visible help articles and runbooks so your buyers self-serve without opening ops tickets.
  • Agent Quality scorecard: drift status (in policy / watch / breach), sanitized flagged conversations, and continuity attestations buyers can audit.
  • Monthly Proof of Operations PDF archive—uptime, drift averages, safety blocks, and token spend in one leadership-ready report.
  • Proof of Compliance certificate, identity verification, resilience scorecard, and SLA breach credit visibility.
  • Partner MSP view: multi-tenant portal for partners managing many customer fleets from one branded surface.
Technical details for security teams
  • White-label branding per customer; Agent Quality drift scorecard; monthly Proof of Operations PDF archive.
  • SLA tracker wired to managed continuity; Stripe invoices and plan management in-portal; SLA breach credits ledger for transparent remediation.

Guardrails and evidence security and audit teams expect.

Safety & governance

  • Central safety policies applied consistently across machines and agents.
  • Violations surfaced with evidence your security team can review, not buried in logs.
  • Human-in-the-loop approval for high-risk tool actions and high-risk drift realignments before execution proceeds.
  • Immutable audit trails, continuity attestations, and optional Concordium anchoring—including CIS-8004 agent registry—for SOC 2 and procurement.
Technical details for security teams
  • ClawGuard: policy management, per-machine and per-agent assignments, agent-level tool validation and prompt sanitisation, gateway request/response inspection, violation feed.
  • Safety pending-actions queue with operator approve/deny; workflow-routed approval routing for high-risk actions.
  • CIS-8004 on-chain agent register/verify/revoke; public Agent Card JSON; continuity attestations hash-anchored on Concordium where configured.
  • Operator audit log with full action history; client portal SLA, tickets, Agent Quality, and compliance reporting.

Control spend and stay free of single-vendor lock-in.

Cost & providers

  • Per-customer token budgets with hard caps, pre-request gates, and alerts so autonomous workloads stay profitable.
  • Use OpenAI, Anthropic, Google, AWS, or on-prem models, switch without re-architecting the fleet.
  • Gateway context cache reduces repeated prompt spend when multiple agents share the same machine.
  • Managed keys or bring-your-own with rotation visibility finance and security can audit.
Technical details for security teams
  • Providers: OpenAI, Anthropic, Google Vertex (Gemini), AWS Bedrock, Ollama for private cloud.
  • Pre-request budget checks; usage reporting for managed and BYOK billing tiers in Command Center.
  • Per-machine context-cache hit rate and bytes saved reported via Gateway heartbeat.

Upgrade agents and extend capability without surprise breakage.

Ship with confidence

  • Roll out agent updates through phased validation with automatic rollback when something regresses.
  • Per-customer auto-update policies with maintenance windows so production waves land when your teams expect them.
  • Extend agents with vetted skills—including WASM sandbox tier—quarantine, test, and release before fleet-wide install.
  • Blueprint evaluation pipeline scores agent configs against golden cases before publish.
  • Structured agent blueprints define purpose, tools, memory, and continuity policy (auto-realign levels, approval gates) in versioned form.
Technical details for security teams
  • Release pipeline: dev, then staging, then canary, then wave rollouts with regression gates; SOUL.md, skills, and memory configs versioned per claw type.
  • Skill catalog with quarantine workflow and architect release gate; ClawHub sync; WASM execution tier in catalog metadata.
  • Automated blueprint evaluation before publish; eval-pass snapshots become known-good restore points.
  • Blueprint continuity config: auto-realign L1–L5 ladder, max level, and require-approval-above thresholds per agent.

Prove fit on your infrastructure before a managed retainer.

Evaluate & deploy

  • Download installers for your OS and validate on your VPC or on-prem, no secrets on the marketing site.
  • Day 0: copy-paste bootstrap from Command Center — online in minutes, not days.
  • Client portal install hub issues personalized commands and tracks progress—credentials never appear on this marketing site.
  • Flexible runtime per machine: systemd on bare metal, full Docker stack, or hybrid (recommended) without re-architecting the fleet.
  • Optional sovereign node: dedicated mesh on your premises or customer-owned cloud so agents stay on your network boundary.
  • Personalized onboarding and credentials issued only through Command Center when you are ready.
  • Same platform operators use in live engagements, so evaluation matches production reality.
Technical details for security teams
  • Debian, Ubuntu, Fedora, plus community portable Linux, macOS, and Windows packages from threesixty.co/claw-tools/latest.
  • Operators and clients mint Tailscale keys and personalized install commands in the portal; install progress is tracked until the machine heartbeats.

Download installers

See it in your environment

Book a live demo or start with a 48-hour AI Health Audit we will map your agent footprint to the platform capabilities you need first.

Book a live demo