ClickNest a tiny self-hosted, AI native web analytics in Go

Name: ClickNest a tiny self-hosted, AI native web analytics in Go
Availability: InStock
Author: ethantheswe

by ethantheswe·Feb 26, 2026·6 points·1 comment

Visit Project View on HN

AI Analysis

●●●BangerShip ItSolve My ProblemDark Horse

LLM auto-names click events from DOM—no manual instrumentation needed.

Strengths

•Single Go binary, zero external dependencies, deploys free to Fly.io
•LLM event naming reads DOM/source to generate 'Clicked checkout button' instead of raw selectors
•Full feature set (funnels, heatmaps, retention, feature flags) in a lightweight self-hosted tool

Weaknesses

•Competes with Plausible, Fathom, Posthog—LLM naming is novel but not a moat
•No clear story on LLM costs at scale or privacy implications of sending DOM to external API

Post Description

I've been building a bunch of random side projects, as I'm sure we all are. Most analytics tools were either too heavy or required juggling multiple free tiers across different subdomains.

After looking what's out there I couldn't find a lightweight analytics platform that worked on the free tier of fly.io.

It's a single Go binary using DuckDB for events and SQLite for metadata, so there are no external dependencies to manage. Drop a script tag on your site and it auto-captures pageviews and clicks.

The part I'm most excited about: it uses an LLM to automatically name raw click events into something human-readable. Instead of seeing button.btn-primary#checkout in your dashboard, you see "Clicked checkout button". No manual event instrumentation needed.

One fly launch and you have your own analytics instance for free.

GitHub: https://github.com/danielthedm/clicknest