FAQ
Frequently asked questions about GridWork HQ.
FAQ
General
What is GridWork HQ?
GridWork HQ is a self-hosted agency operations dashboard with 17 AI pipelines, 12 automated cron jobs, and a git-backed knowledge vault. It is built for solo agency owners who want AI-powered automation without giving up control of their data.
How much does it cost?
GridWork HQ is a one-time purchase of $199 from gridwork.dev. There are no subscriptions or recurring fees for the software itself. You pay separately for hosting (optional) and API usage (Anthropic, etc.).
What AI provider does it use?
GridWork HQ is powered by Claude (Anthropic). It uses three model tiers — Haiku for lightweight tasks, Sonnet for implementation and drafting, and Opus for planning and review. The pipeline definitions specify which model to use automatically.
Can I use it with OpenAI or other providers?
No. GridWork HQ is built specifically for Claude via the Claude Code CLI. The pipeline system depends on Claude Code's tool-use capabilities.
Do I need to know how to code?
Basic familiarity with the command line is required for installation and configuration. The setup wizard handles most configuration interactively. You do not need to modify code for normal operations.
Installation
What are the minimum system requirements?
Node.js 20+, 8 GB RAM, 10 GB disk, macOS 13+ or Ubuntu 22.04+. See Prerequisites for the full list.
Can I run it on Windows?
GridWork HQ is tested on macOS and Linux. Windows is not officially supported. If you need Windows, consider running it in WSL2 (Windows Subsystem for Linux) or Docker.
Do both packages need to run on the same machine?
No. The dashboard and pipeline server communicate over HTTP. You can run the dashboard on Vercel and the pipeline server on your local machine, or split them across different servers.
Is Docker required?
No. Docker is one of four deployment options. You can also run locally, on a VPS, or on Vercel + Tailscale. Docker support was added in v1.2.0.
Pipelines
How many pipelines are included?
17 AI pipelines covering lead generation, client delivery, and operations. Plus a document compiler with 7 compile types.
Can I add my own pipelines?
Yes. Pipelines are Markdown files in the knowledge vault. Create a new .md file in knowledge/system/rules/pipelines/, restart the pipeline server, and it is available immediately. See Creating New Templates.
How much do pipeline runs cost?
It depends on the model tier and the complexity of the task. A Haiku maintenance task costs a few cents. An Opus planning pipeline with multiple passes can cost a few dollars. Check your Anthropic dashboard for usage.
Can I run pipelines without the dashboard?
Yes. You can trigger pipelines via the pipeline server API directly:
curl -X POST http://localhost:8750/pipelines/run \
-H "Authorization: Bearer YOUR_TOKEN" \
-H "Content-Type: application/json" \
-d '{"pipeline": "prospect", "args": "barbershops in Atlanta"}'Data and Privacy
Where is my data stored?
All data stays on your infrastructure. The dashboard uses a local SQLite database (or Turso if on Vercel). The knowledge vault is a git repository on your machine. Pipeline outputs are saved to your local file system.
Does GridWork Digital have access to my data?
No. GridWork HQ runs entirely on your infrastructure. GridWork Digital has no access to your instance, data, or API keys.
Is my data sent to third parties?
Pipeline runs send prompts to Anthropic's API. If you configure Notion, Stripe, or other integrations, data is sent to those services as part of normal operations. No data is sent to GridWork Digital.
Integrations
Which integrations are required?
Anthropic (for AI) and GitHub (for authentication) are required. Notion is strongly recommended for CRM functionality. All other integrations are optional.
What happens if I skip an optional integration?
The dashboard shows "Not configured" for features that depend on missing integrations. The app never crashes on missing API keys.
Can I add integrations later?
Yes. Edit your .env.local file with the new API keys and restart the dashboard. No re-running of the setup wizard is required.