// Platform / Application infrastructure control plane

One operating layer for the infrastructure your applications need.

AgentDeploy starts after a deployable artifact exists. It gives your agent a consistent way to prepare releases, place workloads, connect services and operate applications across infrastructure you choose.

AgentDeploy is a neutral control plane. Infrastructure capacity remains supplied by your cloud, runtime, network and SaaS providers or by systems you operate.

application lifecyclecontrolled workflow
01 / releaseVerify the artifactContent-addressed version and release context
02 / deployPlace the workloadPreview, production and eligible runtime target
03 / connectPublish the right pathGateway, Mesh or Tunnel with explicit policy
04 / operateObserve and change safelyHealth, activity, rollback and audit context
A single application model carries the release from an agent workspace into the infrastructure and access boundary you approve.

// From one application to a portfolio

Use the same lifecycle as your application count grows.

The control plane keeps releases, environments, runtime placement, service entry points and operational state together, so each new application does not require a new collection of dashboards and scripts.

Available foundation

Release and deployment control

Package verified artifacts, deploy separate previews, approve production changes and keep rollback identities explicit.

  • Static, Worker and OCI workload profiles
  • Immutable releases and deployment generations
  • Health-aware application and exposure state
Available foundation

Controlled service connectivity

Give applications stable public or private paths while keeping identity, network and source-service boundaries visible.

  • Gateway / Mesh / Tunnel
  • Managed TLS and custom domains where configured
  • Temporary, scoped access for agent tasks
Integration path

Provider and SaaS adapters

Use provider MCP servers, CLIs and APIs through reviewed Skills that guide local authorization and execution.

  • Cloud and application runtime providers
  • Identity, network, CDN and observability systems
  • Cost, placement and migration workflows
Planned ecosystem

Open customer-side runtime

Customer-side Edge Runtime, clients and adapters are planned as open source so teams can inspect their behavior and trust boundaries.

  • Cloud, edge, homelab and private data center targets
  • Inspectable update and connectivity behavior
  • A portable path beyond any single provider

// Provider ecosystem

Unify control without pretending every provider is the same.

AgentDeploy keeps provider-specific capabilities and ownership intact, then gives agents a common operating path for the decisions that can be normalized safely.

Available foundation

Connected runtime targets

Registered nodes and compatible provider runtimes expose their actual capabilities to deployment planning.

AWS / GCP / Azure / edge / on-prem
Integration path

Existing provider tools

Skills can coordinate provider MCP servers, CLIs and APIs while authorization remains local and reviewable.

Cloudflare / Vercel / Lovable / v0 / provider CLI
Planned ecosystem

Infrastructure SaaS

Identity, private networking, delivery and security services can join the same application operating context over time.

Okta / Tailscale / Cloudflare / Akamai / more

Named services describe intended integration categories or compatible provider paths. Current availability must be checked through live platform capabilities and documentation.

// Clear responsibility boundaries

Automation works better when ownership stays explicit.

The agent can prepare and execute routine infrastructure work. People still approve consequential changes, and infrastructure providers continue operating the capacity and services they supply.

01 / AGENT

Plan and execute

Inspect requirements, prepare provider-specific steps, package releases, deploy previews and report observed state.

02 / CUSTOMER

Choose and authorize

Select infrastructure, grant scoped access, define security boundaries and approve production or sensitive changes.

03 / PROVIDER

Supply and operate capacity

Cloud, runtime, network and SaaS providers continue delivering their services under their own operating models.

// Next step

Start with one application and the infrastructure you already have.

Connect an agent, identify an eligible target and run the first release through a separate preview.