Loading case study…
Loading case study…
Ready when you are.
Start a conversationWork/AI
Live
Open-source AI principal: a person's front door for inbound requests. Humans and agents pitch you; Gatekeeper judges against your plain-English rules, blocks spam, and only delivers double-opt-in intros — with a public ledger as its reputation.

Open-source AI principal: a person's front door for inbound requests. Humans and agents pitch you; Gatekeeper judges against your plain-English rules, blocks spam, and only delivers double-opt-in intros — with a public ledger as its reputation.
Problem
Inbound is noisy. Founders and hiring managers get cold pitches from humans and AI agents with no shared reputation layer. Saying yes/no by hand doesn't scale; a silent auto-accept path invites spam and bad fits.
Solution
Anyone who wants to reach you sends a request to your Gatekeeper. An embedding + LLM judge scores relevance against your published gate rules, rejects spam/deal-breakers (with a public reason and suggested rewrite), and only advances accept-path requests into a double-opt-in intro. A public ledger (requests, accept rate, spam blocked, helpful ratings) is the principal's reputation — inspired by the “Principals” framing that agents are untrusted until they earn trust.
Indie SaaS founders and remote tech hiring worldwide who want an AI front door that can say no — without hiding how it behaves.
Live in production
AGPL-3.0. Public GitHub — https://github.com/arsalan-arshad/gatekeeper. Hosted at https://gatekeeper.arsalandev.studio/. Contributions welcome.
Want something like this built for you?
Start a conversation — sole builder, daily progress, fixed-scope offers.
Start conversation