PRD

Document ID: PRD-001 Version: 1.0.1 Status: Draft — this document is the Day-1 product vision, unrevised since project inception. It has not been reconciled with ADR-0010 (2026-07-06), which ratified Path A: GA ships as a single-node appliance, not the multi-service, horizontally-scaled platform implied below. Current reality vs. §2 “Primary Technologies” (corrected 2026-07-07; see the README’s architecture section for the kept-current architecture description): Rust and Angular are real. NestJS is not built — the dashboard SPA talks directly to wovyr-server; the NestJS BFF is explicitly deferred until production auth needs it (dashboard overview). NATS is not built — there is no message broker; wovyr-events is a custom in-process event/webhook system (see ADR-0005’s current-status note). PostgreSQL, Redis, and Qdrant are all real but optional, feature-gated backends (postgres/redis/tiered-memory cargo features) — the shipping single-node binary’s default is file-based storage under ~/.wovyr, not a hard dependency on any of the three. Kubernetes: a real, offline-validated Helm chart exists for the single-binary topology (deployment/helm/wovyr/, replicas: 1 by product decision), not the multi-replica HA cluster this vision implies, and it has never been applied to a live cluster. Docker is real (deployment/docker-compose.yml). Anything described here that isn’t built and isn’t yet tracked in 18-roadmap/v1.0/ (GA) or 18-roadmap/future/ (post-GA) has been added to the latter as of this revision — see §25 below. Positioning note (2026-07-15): the product framing below also predates ADR-0011’s repositioning — the current product is the Generative UI Trust Runtime (PRD-005, shipped v1.2 scope) with this platform as its engine; see the updated vision. Owner: Wovyr AI Platform Team Last Updated: 2026-07-15

1. Purpose

This Product Requirements Document (PRD) defines the product vision, scope, objectives, high-level requirements, and release strategy for the Wovyr AI Platform.

It serves as the authoritative reference for product planning and aligns engineering, architecture, documentation, testing, and roadmap decisions.

Detailed functional specifications are maintained in companion documents.

2. Product Overview

Product Name

Wovyr AI Platform

Product Category

Enterprise AI Platform

Product Type

Open-source infrastructure platform for building, deploying, and operating intelligent autonomous systems.

Primary Technologies

Rust
Angular
NestJS
PostgreSQL
Redis
Qdrant
NATS
Kubernetes
Docker

3. Vision

Enable developers and organizations to build production-ready AI applications using a unified platform rather than assembling numerous disconnected technologies.

4. Mission

Deliver a modular, secure, extensible, cloud-native platform for AI workflows, agents, memory, plugins, and enterprise operations.

5. Problem Statement

Organizations currently face several challenges when building AI-powered systems:

Fragmented tooling
Vendor lock-in
Operational complexity
Weak observability
Limited workflow durability
Inconsistent security
Poor extensibility
High maintenance costs

Wovyr AI Platform addresses these challenges through a cohesive, modular architecture.

6. Product Goals

The platform aims to provide:

AI Runtime
Workflow Engine
Memory Engine
LLM Gateway
Tool Runtime
Plugin SDK
Dashboard
CLI
SDKs
Enterprise APIs

7. Business Goals

The platform seeks to:

Reduce AI infrastructure complexity.
Increase developer productivity.
Enable enterprise adoption.
Foster a sustainable plugin ecosystem.
Maintain provider independence.

Refer to business-goals.md for detailed objectives.

8. Target Users

Primary user groups include:

Individual developers
AI engineers
Platform teams
Startups
Enterprises
Researchers
Systems integrators

Detailed personas are defined in personas.md.

9. Product Scope

In Scope

AI runtime
Workflow orchestration
Semantic memory
Multi-provider LLM support
Tool execution
Plugin framework
Visual dashboard
REST APIs
gRPC APIs
CLI
Cloud-native deployment

Out of Scope

Training foundation models
Proprietary AI hosting
Consumer chat applications
Low-code website builders
General-purpose database replacement

10. Product Pillars

The platform is organized into the following strategic pillars.

AI Runtime

Responsible for reasoning, planning, execution, context management, and multi-agent orchestration.

Workflow Engine

Responsible for durable execution, scheduling, checkpoints, retries, compensation, and event-driven workflows.

Memory Engine

Responsible for semantic memory, episodic memory, retrieval, ranking, embeddings, and knowledge management.

LLM Gateway

Provides a unified abstraction for multiple AI providers.

Tool Runtime

Executes platform tools with controlled permissions and isolation.

Plugin Framework

Allows extensions without modifying the platform core.

Dashboard

Provides operational visibility and administrative capabilities.

11. Core Capabilities

The first major release targets:

Agent creation
Workflow authoring
Workflow execution
Tool execution
Memory retrieval
Multi-provider LLM routing
User management
Project management
Plugin installation
Observability

12. Functional Overview

High-level capabilities include:

Create AI agents
Execute workflows
Invoke tools
Store and retrieve memories
Manage prompts
Configure providers
Monitor executions
Review logs
Configure plugins
Manage users

Detailed requirements are documented in functional-requirements.md.

13. Non-Functional Requirements

The platform should emphasize:

Performance
Reliability
Security
Scalability
Maintainability
Extensibility
Portability
Testability
Observability

Detailed requirements are maintained in non-functional-requirements.md.

14. Product Architecture

The platform consists of the following major domains:

Runtime
Workflow
Memory
LLM Gateway
Tool Runtime
Plugin SDK
Dashboard
API
CLI
Infrastructure

Detailed architecture is documented under docs/02-architecture/.

15. Deployment Targets

Supported environments:

Local development
Docker
Kubernetes
Private cloud
Public cloud
Hybrid cloud

16. Success Criteria

The product is considered successful when it:

Supports production AI workloads.
Demonstrates stable public APIs.
Enables provider independence.
Provides durable workflow execution.
Encourages community contributions.
Supports enterprise deployments.

Detailed KPIs are defined in success-metrics.md.

17. Release Strategy

Phase 1

Documentation and architecture.

Phase 2

Core runtime and workflow engine.

Phase 3

Memory engine and LLM gateway.

Phase 4

Plugin SDK and tool runtime.

Phase 5

Dashboard and APIs.

Phase 6

Distributed execution and enterprise features.

18. Risks

Major risks include:

Rapid AI ecosystem evolution.
Scope expansion.
Integration complexity.
Balancing flexibility with simplicity.
Maintaining API stability.

Risk management details are maintained separately.

19. Assumptions

The PRD assumes:

Rust remains the primary implementation language for core services.
AI providers will continue evolving behind stable abstractions.
Kubernetes remains a primary deployment target.
Plugin ecosystems continue to grow.
Enterprises require strong governance and observability.

20. Constraints

Current constraints include:

Open-source licensing.
Multi-platform compatibility.
Cloud neutrality.
Modular architecture.
Stable public interfaces.

21. Dependencies

The platform depends on:

Rust ecosystem
PostgreSQL
Redis
Qdrant
NATS
Docker
Kubernetes
OpenTelemetry

Implementations should isolate these dependencies through abstraction layers wherever practical.

22. Companion Documents

This PRD references the following documents:

Vision
Mission
Business Goals
Success Metrics
Product Scope
User Personas
User Stories
Functional Requirements
Non-Functional Requirements
Acceptance Criteria
Roadmap
Architecture Overview
ADRs

Together, these documents constitute the complete product specification.

23. Traceability

Every implementation artifact should be traceable to this PRD.

Traceability includes:

Architecture decisions
Rust crates
APIs
Database schemas
UI features
Test cases
Documentation
Release milestones

Maintaining traceability ensures that engineering work remains aligned with product objectives.

24. Approval

This document should be reviewed by:

Product Management
Solution Architecture
Engineering Leadership
Security
Developer Experience
Documentation Team

Approval indicates alignment on product direction prior to implementation.

25. Technology Gaps Tracked for Future Versions

Added 2026-07-07, alongside the header’s divergence note. Three technologies named in §2 have no implementation and, until now, no tracked future work item either — captured here so they are deferred deliberately, not silently dropped:

NATS-backed distributed event bus — a real cross-replica message bus is only meaningful once the platform runs more than one replica (v1.1 “Scale-Out,” ADR-0010). Tracked as ticket DIST-B9 in phase3-scale-distribution-tickets.md Track B.
gRPC / broader protocol interop — the shipping API is REST/JSON + SSE only. Tracked under FUT-005 (“Ecosystem & Interoperability”), 18-roadmap/future/B5-ecosystem-interop.md, alongside the existing MCP-gateway direction.
S3-compatible object storage — nothing in the codebase uses object storage; plugin packages are local content-addressed files and wovyr admin backup writes to a local path. Tracked as a remote-destination option under GA-002, 18-roadmap/v1.0/A2-reliability-ha-dr.md §4.1.

26. Revision History

Version	Date	Description
1.0.2	2026-07-15	Added a positioning note: the framing predates ADR-0011’s Generative-UI-Trust-Runtime repositioning (PRD-005, shipped v1.2 scope); no content changed
1.0.1	2026-07-07	Added a header divergence note correcting §2’s tech-stack claims against ADR-0010/reality (NestJS and NATS were never built; Postgres/Redis/Qdrant are optional feature-gated backends, not primary dependencies), plus new §25 tracking the three technologies (NATS distributed event bus, gRPC, S3 object storage) that had no implementation and no tracked future work item until now. Found during a project-wide doc review; no product scope changed
1.0.0	2026-06-26	Initial master PRD