Marketplace / Blueprint 02
LIVEAI Experience / intelligence

Private AI Workspace

One private workspace for models, knowledge, tools, search, voice and collaborative AI.

Review my build

Interactive platform map

Architecture in context

The focused blueprint, its required foundation, and declared recommendations.

Core Selected Automatic Required path
Blueprint 02 / AI Experience

Private AI Workspace

One private workspace for models, knowledge, tools, search, voice and collaborative AI.

2 vCPU4 GB RAM7 services
01 / Problem

What this replaces

Users bounce between separate AI subscriptions for chat, search, files, coding, images, voice and prompts, fragmenting data and governance.

02 / Outcome

What your team gains

A single authenticated AI workspace that can talk to multiple models and attach organizational knowledge, tools and controlled external services.

03 / Capability

What is inside the blueprint

Multi-model and side-by-side chat
Files and image uploads
Web search with source retrieval
Browser/Python code execution options
Cross-conversation memory
Voice input/output and voice/video workflows
Image generation/edit integrations
Automations and recurring prompts
Structured task management
Knowledge bases with vector or full-context modes
Hybrid BM25 + vector retrieval with reranking
Agentic document retrieval
Model presets with instructions, tools and knowledge
Notes and AI-assisted editing
Shared channels with model tagging
MCP, OpenAPI, Python tools, pipelines, skills and prompts
RBAC, groups, per-resource permissions, SSO/OIDC/LDAP, SCIM and API keys
Usage analytics, model evaluation/A-B testing, banners and webhooks
OpenTelemetry and horizontal scaling options
04 / Architecture

How it fits the platform

Authentik -> Open WebUI -> LiteLLM/local model endpoints + SearXNG + knowledge/pgvector + MCP/OpenAPI tools + speech/image services.

Included services

Open WebUI, LiteLLM, Authentik, SearXNG, PostgreSQL/pgvector, Kokoro TTS, Gemini TTS proxy

Platform requirements
  • AICORTEX Core PlatformThe workspace's containers, routing, database and boot protection are provided by the Core substrate.
05 / Delivery

From prerequisites to operation

Prerequisites
  1. AICORTEX Core
  2. At least one configured model endpoint/provider
  3. User identity strategy
  4. Optional knowledge collections and tool servers
Deployment
  1. Deploy Open WebUI against PostgreSQL/pgvector
  2. Connect LiteLLM and local model endpoints
  3. Integrate Authentik
  4. Configure web search and speech
  5. Register tools/MCP/OpenAPI endpoints
  6. Define roles/groups and workspace defaults
Configuration
  1. Available models and model presets
  2. Knowledge and retrieval defaults
  3. Web search policies
  4. Voice/image providers
  5. Role/group access
  6. API keys and automation permissions
  7. Analytics and evaluation settings
Operations
  1. User lifecycle
  2. Database backup
  3. Provider health
  4. Extension compatibility
  5. Usage/cost monitoring
  6. Permission review
06 / Combinations

What this unlocks with other layers

Private AI Workspace + Universal AI Gateway

Private AI + Model Freedom

Users get one interface while builders can change providers underneath it without changing the experience.

Private AI Workspace + Voice & Multimodal Experience + Real-Time & Mobile Experiences

Multi-Channel Intelligence

The intelligence layer is reusable across interfaces rather than tied to one chat surface.

07 / Technology

Technology behind this capability

Open WebUIRUNNING - v0.11.0 in censusLiteLLMRUNNINGSearXNGRUNNINGKokoro 82MRUNNINGGemini TTSRUNNING through proxypgvectorRUNNING in enterprise DB imageAuthentikRUNNING - server + worker + DB + Redis