Mike Lay
Mike Lay is an independent software engineer and creator based in San Francisco. His public work focuses on open-source developer tools, AI/ML projects, and infrastructure that lets assistants and agents work with SaaS primitives.
At a Glance
- AI-agent and assistant developers
- Open-source and self-hosting developers
- Teams building internal SaaS-style workflows
- GitHub development teams
- +1 more
AI Tools by Mike Lay
(1)The Last SaaS
Headless SaaS Backend for AI Agents
Discussions
No discussions yet
Be the first to start a discussion about Mike Lay
Latest News
Products & Services
An open-source, headless SaaS backend and hosted service for AI assistants and agents. It provides records, files, team accounts, permissions, notifications, and audit logs through an HTTP API, MCP, and CLI; it can run on Bun with SQLite or PostgreSQL and can be self-hosted.
A GitHub Action that turns commits and other comments into poetry and can post a poem to a pull request. It uses a small fine-tuned 4-bit model; the associated Q4_K_M GGUF is based on Qwen3-0.6B and is intended for CPU inference with llama.cpp.
An open-source Python library for downloading closed captions/subtitles from YouTube.
A 4-bit quantized 0.8B-parameter text-generation model published by Mike Lay on Hugging Face to power PR Poet; the model page lists a 484 MB GGUF for CPU inference.
Market Position
The Last SaaS is positioned as a headless, agent-first SaaS foundation: its differentiator is bundling common SaaS primitives—records, files, accounts, permissions, notifications, and audit logs—behind MCP, an API, and a CLI, while remaining open source and self-hostable. PR Poet occupies a lightweight developer-automation niche by using a small fine-tuned model rather than requiring a large hosted model.
Leadership
Founders
Mike Lay
Independent software engineer and product builder; GitHub identifies him as Mike Lay (mkly), based in San Francisco. His public work includes open-source Python, TypeScript, GitHub Action, and AI/ML projects. ORCID records a 2025 Stanford Open Source Software Prize distinction.
Executive Team
Mike Lay
Independent software engineer and creator
San Francisco-based software engineer whose public projects span open-source web infrastructure, Python/TypeScript tooling, GitHub Actions, and small AI/ML models.
Founding Story
The Last SaaS began as an experiment to find the primitives needed by a minimal SaaS backend and let Claude, ChatGPT, and other assistants handle the rest. The project is presented as open source with a free hosted version and was created by Mike Lay.
Business Model
Revenue Model
Open-source/self-hosted software with a hosted version of The Last SaaS offered free on the project website.
Pricing Tiers
The Last SaaS website states that the hosted version is free.
Target Markets
- AI-agent and assistant developers
- Open-source and self-hosting developers
- Teams building internal SaaS-style workflows
- GitHub development teams
- Developers experimenting with compact, locally runnable language models
- AI assistants and agents that need structured records, files, permissions, reminders, and audit trails
- Developers building headless or agent-facing SaaS applications
- Self-hosted internal tools and lightweight team workspaces
- GitHub teams that want automatic poetic summaries of pull-request commits
- Developers needing a small CPU-runnable text-generation model
- Programs that need to download YouTube subtitles/captions