Skip to main content

Valkyr Labs Docs

Valkyr spec-to-secure-running-stack architecture connecting a business schema to generated application layers, governed memory, security, and deployable infrastructure

Build and run AI software your business owns.

Valkyr is spec-to-stack, not spec-to-SDK. It continuously composes and delivers the owned application through CG/CI/CD, then runs governed agentic operations on that stack. Business schema, realtime data, GrayMatter memory, deterministic workflows, enterprise code, security, encryption, and deployment remain one operating system.

This docs portal is focused on what is live in the product right now, how the layers fit together, and how to run the resulting software on hosted, private, or fully offline infrastructure.

Start here: ValkyrAI Now Shipping

Quick Start by Goal

GoalRead ThisThen Do This
Model a real business domainUnderstanding OpenAPI SpecificationsDefine the entities, rules, permissions, and operations that become the application contract
Generate an owned composable stackCreating Your First ValkyrAI ProjectGenerate the enterprise backend, data model, APIs, security, clients, and application surfaces from the spec
Run agentic operationsWorkflow Studio (Production Guide)Compose portable, deterministic visual workflows in /workflow/builder
Build APIs from specsOpenAPI Design CenterOpen /openapi-designer
Run AI automation with guardrailsSageChat + Command SystemOpen dashboard and launch SageChat
Control memory + token economicsGrayMatter Memory + CreditsOpen /graymatter-memory and /credits/:accountId
Operate memory systems directlyGrayMatter Product DocsTune memory actions and spend posture
Deploy hosted, private, or offlineDeployment OptionsChoose the infrastructure boundary without giving up the generated codebase
Configure integrations and publish toolsIntegration Accounts + MCP MarketplaceOpen /integration-accounts and /mcp/marketplace
Tune onboarding for your business modelOnboarding JourneysOpen /onboarding

Product Stack

  • GrayMatter: business knowledge graph, durable shared agent memory, invariants, retrieval receipts, and governed context associated with live domain data.
  • ThorAPI: deterministic spec-to-stack generation and CG/CI/CD for composable enterprise Java applications, data models, APIs, security, clients, UI surfaces, and deployment.
  • ValkyrAI: portable, n8n-style deterministic visual workflows for governed agentic operations across models, tools, and business processes.
  • SageChat and ValorIDE: conversational and developer control surfaces for the same application system.
  • TrustFabric: encryption, RBAC, receipts, auditability, tenant boundaries, and compliance evidence.
  • Hosted, private, and offline runtime: tenant-isolated cloud scale, private deployment, embedded operation, or complete operation without internet access.
  • MCP Marketplace: publish, discover, and monetize protocol-ready services.

Core Architecture Promise

Valkyr is designed to remove the handoff gap between building software and running the business:

  1. The business schema remains the source contract instead of disappearing into prompts.
  2. GrayMatter associates agent memory and decisions with the domain graph and realtime business data.
  3. ThorAPI deterministically composes the complete enterprise application stack from the specification.
  4. CG/CI/CD continuously regenerates, integrates, tests, and delivers the application as the domain evolves.
  5. ValkyrAI keeps operational logic in portable, deterministic visual workflows instead of stochastic agent-generated code.
  6. GrayMatter and ValkyrAI run governed agentic operations against the same business graph and realtime data.
  7. The resulting code, workflows, memory, and deployment can be inspected, customized, operated, and owned by the customer.
  8. The stack can run tenant-isolated, privately, embedded, air-gapped, or fully offline.

Ideal Customer Profile

Valkyr is designed for buyers with budget, authority, operational pain, and a real system to build or run:

  • funded or revenue-producing product teams and software vendors;
  • agencies, consultancies, systems integrators, and fractional CTOs seeking repeatable delivery leverage;
  • enterprise and regulated teams with governance, modernization, encryption, audit, tenant-isolation, or private-deployment needs;
  • funded government, nonprofit, research, robotics, and embedded programs with real workflow or restricted-network requirements;
  • serious founders and solopreneurs with funding or revenue, a concrete operating problem, and urgency.

Strong first engagements begin with an existing process, schema, or dataset; a named decision owner; an approved budget and deadline; and a bounded paid architecture or implementation sprint.

What Is Different About This Documentation

  • It is route-aware: every guide maps to real product routes.
  • It is ops-aware: endpoints, failure modes, and guardrails are documented.
  • It is handoff-friendly: each page includes practical steps for new teammates.

Need Help