Planned platform
A specialized AI workspace for infrastructure decisions and operations
Service Planet is planned as a DevOps and IT infrastructure team in one platform for internal IT teams and MSPs. Its own AI assistant would help discover and assess environments, compare proposed changes, prepare qualified actions and follow verified outcomes. Chat, topology, cost and status stay central; deeper inventory and administration open when needed, without hiding required scope, risks or approvals.
Inventory, assessment and topology in context
Inventory, assessments and current topology would show what is known, how it was collected and where coverage is incomplete. Observed relationships stay distinct from inferences; actual charges stay distinct from proposed topology and estimated costs. Configurable monitoring and findings would use Security, Performance and Health & Availability categories. Status must distinguish up, degraded, down and unknown, with freshness visible: a reachable resource is not necessarily healthy, and stale or missing data cannot prove health.
Your workspace, on a portable foundation
Service Planet’s own web, iOS and Android management clients are the planned primary experience, separate from the deployable backend/control plane and OS-installed server agents. Each client would use the same governed APIs, authentication, organization and deployment permissions, approvals and audit. Model Context Protocol (MCP) access for individually qualified external AI clients is optional, not a prerequisite.
Your infrastructure-focused AI assistant
The specialized assistant is intended to support Docker, Kubernetes, Windows, Linux, servers, applications, databases and networking, including load balancers and firewalls. Model sourcing remains under evaluation: adapting an existing model or using suitable open-source/open-weight components does not mean training a foundation model from scratch. No model, training approach or license is selected. Specialization does not guarantee correctness or safety; software-enforced permissions, validated tools, policy gates, sandbox evaluation and audit remain essential.
Planned technical research would identify authoritative sources, relevant versions and freshness, exposing uncertainty and checking compatibility, security and lifecycle status. Patch and update management connects discovery and advisories to applicability and dependency review, then to separately approved, qualified maintenance. Research or a recommendation never authorizes an upgrade.
Who it’s for
One organization may have several deployments across mixed cloud and dedicated infrastructure. Internal IT teams and MSPs would use the same planned capabilities with access confined to the organizations, deployments and data authorized for each task.
For MSPs
Keep each customer’s inventory, topology, cost and evidence separate. Prepare technical and business views from the same scoped snapshot, and retain customer boundaries through any approved action and its results.
For internal IT teams
Assess your organization’s deployments, compare proposed changes and plan qualified operations. Keep current topology, cost basis, operational findings and the evidence behind each decision within view.
The dedicated internal IT page is being prepared.
The intended commercial model is open-source self-hosted software, optional managed hosting and separately purchasable support. The product is not released; license, prices, support terms, SLA and availability remain undecided.
A complete lifecycle, delivered in qualified stages
The full plan spans broad AWS, Azure and Google Cloud services, productivity platforms and dedicated environments. A read-only workspace and collector agents are the recommended starting point, not the limit of the product. Later management, patching, private-cloud provisioning and migration require exact approved plans, fresh-state checks, verified outcomes and action-specific recovery. Optional self-healing adds an on/off control for selected resources or domains; the recommended default-off design permits only separately authorized repair runbooks. Monitoring signals, cadence and retention need definition; notification channels and alert settings remain tentative.