Tools, with their status made explicit.

Runs language, vision, and embedding models locally.

Private model work, retrieval embeddings, prompt development, and local fallbacks.

One OpenAI-compatible gateway in front of local and cloud models.

Keeps applications independent from any single model provider and gives each consumer one stable API.

Node-based local image and video generation.

The controllable generation engine behind visual experiments, nex_img, and production artwork.

A studio-facing generation layer around local visual models.

Turns repeatable generation workflows into a cleaner API for brand imagery and reviewed creative output.

A self-hosted chat and tool-use interface for local models.

Conversational retrieval and operator workflows that connect models to search, files, automation, and design tools.

A self-hosted multi-provider chat workspace.

Used to compare a second interface for models, retrieval, and MCP-connected tools without changing the gateway.

A retired personal-assistant service that joined models, retrieval, and operator tools in one interface.

Previously provided one assistant surface and proved a resilient routing pattern; retained here as historical context after the service was retired.

A vector database for semantic search and retrieval.

Stores embedded legal, organizational, and knowledge collections queried through purpose-built tools.

An open-source, API-first CRM.

Custom objects and workflows model funders, stakeholders, attorneys, opportunities, and follow-up as living pipelines.

Dashboards and practical analysis over operational data.

Turns pipeline records into views people can read without hand-writing a new query for every question.

Visual workflow automation with code where the edge cases need it.

Coordinates intake, enrichment, research, notifications, publishing, and human review gates.

A privacy-respecting metasearch engine.

Supplies web evidence to research and enrichment flows without binding the system to one commercial search API.

Automated open-source intelligence collection.

Adds structured public-web evidence to research pipelines where broad discovery matters more than a single lookup.

A unified API for scheduled social publishing.

Explored as the final publishing handoff for generated and human-approved creative, without maintaining one integration per network.

Open-source optical character recognition.

A utility path for turning scanned pages into text before retrieval and document workflows can use them.

A local-first, agent-native design platform.

Provides project workspaces, design systems, editable artifacts, and structured handoffs for agent-authored design.

Origin Creative’s brand-operations layer on top of Open Design.

Applies brand rules, coordinates specialist agents, and routes creative through a human approval gate.

A bridge from agent workflows into native Illustrator documents.

Builds editable files, places artwork, manages layers and swatches, and supports print-production checks.

Agent control for Premiere projects, timelines, effects, and exports.

Installed and exercised as a video-production bridge; it is available when a Premiere editing session calls for it.

An open-source browser-based design and prototyping tool.

Evaluated as an own-your-files collaborative canvas and programmable alternative for interface work.

Agent-native HTML-to-video rendering: write HTML, CSS, and JavaScript, then render a deterministic MP4.

Being adopted to turn web-native compositions and structured data into repeatable rendered video for creative and design pipelines.

A framework for creating videos programmatically with React (source-available; a company license applies to larger organizations).

Being evaluated to build data-driven and templated motion graphics as code, alongside the HTML-to-video path.

A plain-Markdown knowledge vault.

Owns long-lived notes and research in portable files, with selected material available to retrieval.

A structural knowledge graph over source code.

Lets coding agents discover existing symbols, routes, and call paths before they open or duplicate large parts of a repository.

A small set of plain-Markdown stores with distinct jobs.

Separates current code, operating procedures, personal knowledge, and the durable reasons behind decisions.

Self-hosted Git and repository management.

Keeps private source close to the systems it drives while preserving normal branch and review workflows.

A reverse proxy with automatic TLS.

Handles secure web serving and routing for selected applications.

Network DNS with ad and telemetry blocking.

Keeps routine network noise down and supports a cleaner browsing environment.

Self-hosted service monitoring.

Tracks service availability so a silent failure becomes visible.

A content-focused web framework that ships little client JavaScript by default.

Builds this resource site, its case studies, and the Markdown-friendly library.

Static-site hosting at the public edge.

Publishes the committed Astro build after a deliberate review and manual release step.