AIHOT

Reading & Bookmarks

Build a self-hosted industry news site with AI filtering, event clustering and daily briefings.

Use with conditions
Last updated2026-10-03
Open source licenseMIT
PlatformsWeb browser (reader interface) · Docker Compose host · Node.js 24.11+ with PostgreSQL 16 or 17 (alternative)
DeploymentSelf-hosted service

About this project

AIHOT is a self-hosted news-site engine with a reader website and an administration interface. It collects material from configured sources, filters and scores items with language models, summarizes selected material and groups reports into events. Event-level popularity and scheduled daily, weekly and monthly briefings help readers follow a subject without reading every source separately.

The current framework separates site identity, branding and pages in site/ from sources, categories, prompts and selection thresholds in industry/. Owners can adapt these files to a subject and calibrate selection with their own examples. The repository provides 18 demonstration AI sources; the official service's production source list, operating data, model leaderboard, Codex reset monitor and theme-page chronologies are excluded.

Official deployment uses Docker Compose or a documented Node.js and PostgreSQL alternative. The recommended cloud machine has at least two CPU cores and 4 GB of memory. Model API credentials, administrator secrets, recurring service costs, backups and source rights remain the operator's responsibility. Code uses MIT; deployments must use their own name and logo, while bundled fonts and publisher content retain separate terms.

Decision guide
01

Why it is worth a look

  • Configurable source collection and AI filtering
  • Event clustering and popularity ranking
  • Daily, weekly and monthly briefings
02

Before you use it

  • Prepare a server, model API budget, administrator secrets, backups and HTTPS.
  • Use your own name and logo; verify the rights and access conditions of each source.
  • The public engine excludes AIHOT operating data and AI-only modules; review version changes before upgrading.
  • Independent review covers an earlier source snapshot and hosted interfaces; complete deployment of the current engine was not verified.
Manual setupHigh
With an AI agentHigh
DeploymentSelf-hosted service
Additional requirements3 confirmed; others not verified

A machine able to run the documented services; A model API credential and spending budget; Administrator secrets and a backup plan; Your own site name, logo, source list and selection criteria; Confirmed requirements: Command line, An always-on device or service, API key; Some requirements have not been verified. Check the official setup instructions.

03

What it can do

  • Configurable source collection and AI filtering
  • Event clustering and popularity ranking
  • Daily, weekly and monthly briefings
  • RSS, public API and MCP access
  • Administration, selection calibration and service budgets

Who it fits

Good for

Individuals and editorial teams willing to maintain a configurable industry news site

Not a fit

You need a ready-made app without server administration, model fees or source calibration

Getting started

  1. Review the pinned official deployment guide and prepare Docker Compose, or Node.js 24.11+ with PostgreSQL 16 or 17
  2. Clone the official repository, configure SITE_URL, model credentials and administrator secrets locally
  3. Replace the name and logo, select sources and adapt site/ and industry/ to your subject
  4. Build and start the documented services, then inspect content, model budgets and source access before publishing
  5. Configure backups and HTTPS; review migration instructions before each update