openclaw is a standalone TypeScript AI-assistant shell with no natural integration point in the Next.js/FastAPI/Postgres/n8n stack in use here. The closest realistic fit is reading its source for architectural patterns, not depending on or deploying the project directly. Running it as a Proxmox container alongside Ollama is technically possible but adds a large, fast-moving, unfamiliar codebase to a homelab built around small self-authored services.
The 'own your data' local-first positioning echoes life-automation's local-only Vite app and the operator's general preference for self-hosted, non-cloud-dependent tools. But life-automation's companion API is a personal OS dashboard, not a general-purpose AI assistant shell, so the overlap is thematic (local-first AI) rather than a direct product match.
Implementation Options
01Architecture review (research only)
life-automation2-4 hourslow risk
Spend a few hours reading openclaw's source, focused on its context-persistence layer and model-routing abstraction, and write structured notes to inform life-automation's companion API design. No code integration, no new dependency.
+Near-zero risk and cost, directly informs an active design decision
+TypeScript is already familiar from Next.js work, so reading the source is cheap
-Produces no shipped feature by itself
-Value only materializes if the notes actually get used when companion API work starts
02Adopt specific patterns into companion API
life-automation6-10 hoursmedium risk
After the review, port 1-2 concrete ideas (e.g. a structured context-persistence schema, a local Ollama model-routing shim) into life-automation's existing FastAPI/Vite code, written from scratch rather than importing openclaw itself.
+Concrete, testable improvement to the companion API
+Stays inside the existing stack, no new runtime dependency
-Only makes sense once the architecture review is done and companion API work has actually started
-Risk of over-engineering life-automation for scale it doesn't need yet
03Deploy openclaw as a homelab sidecar
life-automation1-2 dayshigh risk
Run openclaw in a container on Proxmox alongside Ollama, exposed via the Cloudflare tunnel, to trial it directly as a personal AI layer.
+Fastest way to see the 'own your data' assistant model working end to end
+Could shortcut the companion API build if openclaw already covers the need
-A repo growing this fast (384k stars since a Nov 2025 creation, 6.8k open issues) usually means breaking changes and thin docs - high maintenance burden for a solo operator running unattended systems
-Adds a large unfamiliar codebase rather than informing a small self-built one
-Duplicates effort with life-automation's own companion API instead of guiding it
If pursued as review-only, the value is a sharper companion API design for life-automation, specifically how to persist personal context locally and route to Ollama, decided before that design gets locked in. It is design input, not a new capability.
life-automation is the one active project explicitly building toward a local-first personal AI layer. Getting the persistence and routing model right early avoids a rework later, and openclaw is a live example of a much larger team solving the same problem.
03 · What Implementation Looks Like Minimal v1 is a research note (not code) added to life-automation's docs summarizing openclaw's context-persistence and model-routing approach, with 2-3 explicit recommendations for the companion API's data model. No files in the running app change yet; the FastAPI companion API is what gets informed once that work actually starts.