Product Strategy designed
AI Cloud is a secure self-service GPU platform for discovering capacity, provisioning compute, accessing running allocations, monitoring usage, and paying based on consumption. The product direction is to evolve from GPU node self-service into an app platform with clear developer, operator, and security surfaces.
Product Goals
| Goal | Meaning |
|---|---|
| Fast time-to-compute | Users should get from intent to usable GPU capacity in minutes |
| Transparent usage and billing | Users and operators can understand cost, credits, burn rate, and billing outcomes |
| Safe multi-user operation | Tenant/project scope, roles, audit, policy, and isolation are first-class |
| Operator-friendly control | Admin and ops teams can manage inventory, releases, incidents, and evidence |
| App platform path | Developers can package apps and use GPUaaS as the runtime/product shell |
Personas
| Persona | Primary need | Portal path |
|---|---|---|
| End user | Launch and operate GPU-backed workloads | Use AI Cloud |
| Tenant/customer admin | Manage projects, access, usage, and billing posture | Use AI Cloud |
| Platform admin | Manage nodes, users, allocations, audit, payments, and operational posture | Operators |
| App developer | Build, package, test, and promote apps on AI Cloud | Build on AI Cloud |
| Security reviewer | Understand controls, gaps, evidence, and release discipline | Security & Production Readiness |
| Architect/engineer | Understand system boundaries, contracts, and implementation model | Architecture |
Current Product Scope
| Area | Included now |
|---|---|
| Auth and access | OIDC-backed auth, role-aware authorization, tenant/project direction, token refresh/logout patterns |
| Catalog and inventory | SKU/catalog availability, node inventory, admin node workflows |
| Allocation lifecycle | Request, provisioning, active, releasing, released, failed, release-failed states |
| Runtime access | Browser terminal, SSH/access posture, terminal token/session flow |
| Billing and payments | Usage accrual, immutable ledger, balance warnings, Stripe checkout/webhooks, payment sessions |
| Storage | Object-storage-backed user storage operations and path safety |
| Admin/Ops | Users, nodes, allocations, audit, payment sessions, operational telemetry direction |
| App platform | App catalog, manifest, SDK, artifact trust, and promotion path direction |
Product Maturity
| Strength | Gap to keep visible |
|---|---|
| Contract-first API/event foundation | Generated API reference needs to become first-class in portal |
| Clear v3 IA and user journey model | UI migration and route consolidation are still active work |
| Billing, allocation, terminal, and storage flows are documented | Operator diagnostics and monetization depth need more product surface |
| App SDK direction is strong | Published SDK artifacts and internal developer onboarding need a formal release path |
| Security and production readiness are explicit | Enforcement separation and environment/ring maturity still need implementation |
Product Operating Principle
The portal should explain the product without becoming the product backlog. Canonical specs, contracts, and implementation plans remain in source docs; the portal gives each team a maintained path through them.