Before Prostir, expertise sits in files, examples, and manual replies. The public landing explains the offer, WebApp guides the creator through building it, and SuperAdmin keeps the service visible.
About Prostir
Turn your expertise into an AI app customers can use.
The story starts when an expert has the knowledge but not the backend: documents, examples, formulas, clients, and a repeatable service. Prostir supplies the product layer around that expertise: identity, knowledge, tools, remote MCP endpoints, quotas, payments, state, memory, and operator visibility.
Ask an AI client
About Prostir
Managed Code foundation
Managed Code turns repeated infrastructure into product primitives: MCP endpoints, knowledge artifacts, and storage contracts the creator does not have to invent.
01
Product surface for the creator
Before Prostir, expertise sits in files, examples, and manual replies. The public landing explains the offer, WebApp guides the creator through building it, and SuperAdmin keeps the service visible.
- Astro landing
- Blazor WebAssembly app
- Blazor Server operator app
Product surface for the creator
Astro landingBlazor WebAssembly appBlazor Server operator app
02
Managed Code foundation
Managed Code turns repeated infrastructure into product primitives: MCP endpoints, knowledge artifacts, and storage contracts the creator does not have to invent.
- ManagedCode.MCPGateway
- ManagedCode.MarkdownLd.Kb
- ManagedCode.Storage
02 Managed Code foundation
ManagedCode.MCPGateway
ManagedCode.MarkdownLd.Kb
ManagedCode.Storage
03
Runtime foundation for customer requests
When a customer asks the app for help, the request does not go straight to a prompt. API, MCP Gateway, Orleans grains, Cosmos persistence, and ManagedCode packages separate creator, app, session, customer, tool, and quota boundaries.
- Orleans owner grains
- Cosmos persistence
- MCP gateway
- Runtime quotas
03 Runtime foundation for customer requests
Orleans owner grainsCosmos persistenceMCP gatewayRuntime quotas
04
Knowledge and search
When the customer asks, files should not become pasted prompt text. Prostir stores sources, builds Markdown-LD/JSON-LD knowledge artifacts, and exposes search/fetch paths so the runtime can answer from approved material.
- Markdown-LD artifacts
- Knowledge search
- Source boundaries
Knowledge and search
Markdown-LD artifactsKnowledge searchSource boundaries
05
Programmable tools for real actions
Some customer scenes need a number, validation, or action. External APIs and JavaScript tools let apps perform calculations and controlled actions, not only answer from documents.
- Jint execution
- Deterministic formulas
- MCP annotations
05 Programmable tools for real actions
Jint execution
Deterministic formulas
MCP annotations
06
State and memory for continuity
When a session has steps, the story needs current state, event, guard, action, and next state. Stateless supports transitions, while Orleans-backed scopes keep current state and user memory inside the app and user boundary.
- Can-fire checks
- Explicit events
- Scoped memory
06 State and memory for continuity
Can-fire checksExplicit eventsScoped memory
07
Commercial layer for paid access
After the preview works, the creator still needs to sell access and protect the runtime. Stripe packages, access rules, and operation quotas turn a useful assistant into a responsible product.
- Stripe packages
- Access grants
- Billable operations
Commercial layer for paid access
Stripe packagesAccess grantsBillable operations
08
Operations model
Public product surfaces show what creators can build today, what needs guided setup, and what is coming next in the product.
- Self-serve builder
- Guided setup
- Product roadmap
08 Operations model
Self-serve builder
Guided setup
Product roadmap
Runtime foundation for customer requests
Read the cases before choosing a build path.
The case studies show how the technical pieces map to real customer-facing AI apps.
01
Product surfaces
The public site explains the offer, the creator workspace builds the app, and the operator console keeps the service visible.
02
Runtime and data
Prostir is built by Managed Code on typed .NET services, Orleans owner-grain boundaries, Azure storage, and ManagedCode packages for MCP, knowledge, and storage. Microsoft.Extensions.AI is the .NET AI abstraction reference for chat clients, embeddings, tools, telemetry, and middleware.
.NET Core runtime and libraries. .NET Aspire Local orchestration and cloud-native app model. Microsoft.Extensions.AI Unified .NET AI abstractions for chat clients, embeddings, function invocation, telemetry, and middleware. Orleans Distributed owner grains for app, user, session, and operation boundaries. EF Core Typed persistence and query model. Azure Cosmos DB .NET SDK Azure NoSQL persistence SDK.
03
Built by Managed Code
Managed Code is not just the footer credit. Prostir uses ManagedCode.MCPGateway, ManagedCode.MarkdownLd.Kb, and ManagedCode.Storage as core backend foundations.
Managed Code Company and engineering team behind Prostir. ManagedCode.MCPGateway Gateway aggregation layer for tools, prompts, resources, and upstream MCP sources. ManagedCode.MarkdownLd.Kb Knowledge/RAG baseline for Markdown-LD and JSON-LD artifacts. ManagedCode.Storage Storage abstraction for files, memory, and runtime artifacts.
04
Agent execution
MCP protocol handling, deterministic JavaScript tools, workflow state, and Stripe billing stay server-side.
About Prostir
Read the cases before choosing a build path.
The case studies show how the technical pieces map to real customer-facing AI apps.