How We Build
We're building AI that thinks like a marketer. Here's a look under the hood at our tech stack, engineering principles, and contributions to the open source community.
Our Tech Stack
Modern tools chosen for speed, reliability, and developer experience.
AI & ML
- Anthropic
- OpenAI
- MCP Servers
- A2A Protocol
- CrewAI
Frontend
- Astro
- Next.js
- TypeScript
- ShadCN
Backend
- Python
- FastAPI
- Node.js
- MongoDB
Infrastructure
- Cloudflare
- Docker
- GitHub Actions
- Terraform
Open Source
We open source the tools we build and use in our own product. If it helps us, it might help you too.
fal-mcp-server
MCP server for Fal.ai image and video generation. Enables AI agents to create visual content.
luminary-mcp-server
Core MCP server powering Luminary Lane's AI marketing capabilities.
ci-runner-setup
Self-hosted CI runner for GitHub Actions + GitLab CI. Cut your CI/CD costs by 90%.
claude-code-principles
Development principles and idioms for building with Claude Code and AI assistants.
More open source projects coming soon
How We Work
Principles that guide our engineering decisions.
Ship Fast, Iterate
We believe in getting things in front of users quickly and improving based on real feedback.
Test Everything
Automated testing and CI/CD ensure we move fast without breaking things.
Design to Falsify
We design experiments that could fail. If you can't prove yourself wrong, you can't learn anything.
Keep It Simple
Occam's Razor guides our architecture. The simplest solution that works is usually the best.
Build Right First
Fast is fine, but accuracy is final. We take time to understand problems before coding.
Know When to Stop
The last 20% costs 80% of the effort. We ship at 80% and iterate, rather than perfecting in isolation.
Want to Build With Us?
We're always looking for talented engineers who are excited about AI and marketing automation.