PatriciAI · lab stack · open tools

The tools behind the useful mischief.

A high-level map of the open-source and open-adjacent software, platforms, and hardware ecosystems in use around my lab, automation, AI, and infrastructure work. This is intentionally not a deployment guide, network diagram, or secrets-adjacent inventory.

local AI automation self-hosting data platforms wearables lab operations
Scope note: this page names the kinds of tools used and links to their public projects or homepages. It avoids fine particulars of how they are deployed here: no private hostnames, credentials, keys, network details, model endpoints, or operational runbooks.

AI, agents, speech, and memory

PatriciAI is built around local-first AI where practical, with OpenAI-compatible interfaces, retrieval, speech, and assistant workflows stitched together under human oversight.

OpenClaw

Personal AI assistant runtime and automation layer used for agent sessions, tools, reminders, memory, and coordination.

GPUStack

GPU model serving and orchestration for local inference, embeddings, reranking, and OpenAI-compatible model access.

LM Studio

Local model exploration and compatibility testing tool; useful for quickly trying models and OpenAI-style workflows.

Speaches

OpenAI-compatible speech service used for local speech-to-text and speech API experiments.

Chatterbox

Open-source text-to-speech stack used for voice generation experiments and assistant voice work.

Flowise

Visual AI workflow builder for prototyping LLM chains, agents, integrations, and retrieval flows.

Automation and integration

The environment favors tools that make repeatable work visible, auditable, and less dependent on memory or heroics.

n8n

Workflow automation platform used for service orchestration, integrations, and business/technical automation experiments.

Home Assistant

Automation and IoT ecosystem for practical environment-aware workflows and device integrations.

ffmpeg

The quiet workhorse behind media conversion, audio/video handling, frame extraction, and automation glue.

Infrastructure and operations

The lab is a proving ground for resilient self-hosting: virtualization, containers, dashboards, reverse proxies, backup posture, and practical service ownership.

Proxmox VE

Open-source virtualization platform for running and managing VM and container infrastructure.

Proxmox Backup Server

Backup platform used as part of the recoverability-first approach to lab and service operations.

Docker

Container runtime and packaging layer for many lab services and experiments.

Nginx Proxy Manager

Friendly reverse-proxy management for routing, certificates, and service exposure.

Homepage

Service dashboard used to keep important systems visible without turning status into a scavenger hunt.

Vaultwarden

Self-hosted Bitwarden-compatible password vault for keeping secrets out of notes, code, and chat.

Gitea

Self-hosted Git service used where local repository hosting and lightweight project control make sense.

Uptime Kuma

Open-source service monitoring and status checks for keeping an eye on what should be alive.

Ubuntu / Debian Linux

The operating-system foundation under much of the lab: boring in the best possible way.

Data, search, and documents

Structured data, document workflows, and retrieval systems matter because agents are only useful when their context is accurate and inspectable.

PostgreSQL

Primary relational database platform for durable app data, structured records, and operational storage.

pgvector

Vector extension for PostgreSQL used for embeddings, semantic retrieval, and memory/search experiments.

Supabase

Open-source backend platform built around PostgreSQL for API, auth, storage, and rapid app work.

NocoDB

Open-source no-code database UI for making structured data easier to view and manage.

MongoDB Community

Document database used where document-shaped application data or compatibility calls for it.

MariaDB

Open-source relational database in the MySQL family, useful for application compatibility and traditional workloads.

MinIO

S3-compatible object storage for self-hosted storage patterns and service integration experiments.

Paperless-ngx

Document management and OCR pipeline for turning piles of paper into searchable operational knowledge.

Stirling PDF

Self-hosted PDF toolbox for conversion, splitting, combining, OCR-adjacent workflows, and document cleanup.

Hardware and embodied tools

The lab is not just servers. Wearables and purpose-built devices are part of the long-term assistant ecosystem: capture, context, voice, ambient memory, and human-centered interaction.

Omi

Open-source AI wearable ecosystem used for exploring always-available memory, capture, and assistant workflows across device types.

rabbit r1

Dedicated AI hardware in the exploration pile. The device is not treated here as fully open-source, but rabbit publishes open-source components and developer-facing material.

Raspberry Pi

Small, flexible utility hardware for lightweight services, edge experiments, device bridges, and lab glue.