Jorvis Capabilities Overview

Last Updated: 2026-03-28 Status: Active overview

This document is a high-level architecture-facing summary of what Jorvis currently exposes.

For current capability truth, use:

docs/FEATURE_MATRIX.md
docs/architecture/MODULE_STATUS_SSOT.md
docs/handoff/CHECKPOINT.md
docs/operations/OPENCLAW_RUNTIME_OPERATIONS.md
docs/architecture/OPENCLAW_MEMORY_ARCHITECTURE.md
docs/configuration/MODEL_POLICY_CURRENT.md

Platform Summary

Jorvis is a data-focused AI platform centered on:

natural-language Text-to-SQL
OpenAI-compatible backend routes
Open WebUI as the canonical chat/auth surface
OpenClaw as an assistant/copilot surface
local Ollama fallback for resiliency and offline-capable paths

Implemented vs Enabled vs Future

Use the categories below when reading capability status:

Category	Meaning
Implemented	Present in code and documented in current repo truth
Enabled	Verified as part of the active production/runtime path
Future	Intended direction or approved architecture, but not yet current runtime truth

Examples:

OpenClaw memory plane: implemented and normalized in current runtime/code truth
OpenClaw admin operator mutations: not delegated to current runtime truth
Observability stack: implemented; production enablement should be verified per deploy
Multi-agent expansion for OpenClaw: future work

Current High-Level Capability Map

Capability Group	Current Status	Canonical Reference
Text-to-SQL	Production	`docs/FEATURE_MATRIX.md`
GraphRAG / schema-aware retrieval	Production	`docs/operations/DEMO_RUNBOOK.md`
Open WebUI integration	Production	`docs/operations/OPENWEBUI_MODEL_SETUP.md`
OpenClaw assistant	Production	`docs/operations/OPENWEBUI_MODEL_SETUP.md`
OpenClaw memory plane	Production	`docs/architecture/OPENCLAW_MEMORY_ARCHITECTURE.md`
Voice gateway	Production	`docs/operations/DEPLOYMENT_RUNBOOK.md`
Monitoring / observability stack	Verified current baseline	`docs/operations/DEPLOYMENT_RUNBOOK.md`

Current Runtime Surfaces

Open WebUI / model surfaces

jorvisauto — primary demo/public model
jorvis_local — local Ollama fallback
jorvis-anydb — script-managed custom OpenAI-compatible registration target
openclaw_agent — canonical OpenClaw WebUI identity

OpenClaw runtime

Canonical OpenClaw current-state references:

docs/operations/OPENCLAW_RUNTIME_OPERATIONS.md
docs/architecture/OPENCLAW_MEMORY_ARCHITECTURE.md
docs/configuration/MODEL_POLICY_CURRENT.md

Current repo-backed OpenClaw truth:

chat primary: google/gemini-3-flash-preview
chat fallback: google/gemini-3.1-pro-preview
memory-search provider: gemini
memory-search model: gemini-embedding-2-preview
memory-search fallback: none
live semantic store: reset/rebuilt on 2026-03-28, intentionally starting from 0 entries after the cutover

For the concrete operations-facing usage model, use:

docs/operations/OPENCLAW_OPS_COPILOT.md

Architectural Nuance

Memory

Do not use historical references to the old MemoryModule as proof of current memory architecture.

Current memory truth for OpenClaw lives under:

analytics-platform/src/internal/openclaw-memory.*

Multi-tenancy

Do not overclaim platform-wide tenant isolation based on historical TenancyModule docs.

Current repo truth is narrower:

tenant-aware scoping exists in the current OpenClaw/session-memory path
older broad TenancyModule activation claims are historical

Non-Canonical Historical Surfaces

These are not current architecture truth:

large historical Open WebUI function bundles
old jorvis-brain model references
Apache AGE-based GraphRAG claims
historical “Semantic Memory removed” / “Multi-Tenancy removed” simplifications

docs/FEATURE_MATRIX.md
docs/architecture/MODULE_STATUS_SSOT.md
docs/operations/OPENWEBUI_MODEL_SETUP.md
docs/operations/DEPLOYMENT_RUNBOOK.md
docs/handoff/CHECKPOINT.md