The exact open-source stack for capturing leads, researching, structuring knowledge, writing, creating voice, and rendering video.
What's inside
Read the first steps here. Unlock the full guide below for the remaining setup and copy-ready material.
You do not need all seven repos installed. Pick the stage where work currently stalls, prove that one, then connect the next stage.
Git and a GitHub account
Claude Code, Codex, Cursor, or another coding agent for setup help
Docker for DeskcommCRM or WeKnora
Python 3.12 for MarkItDown
Node.js 22 and FFmpeg for HyperFrames
A real workflow to test, such as a lead, research brief, document set, script, or video
Pro tip: Start with one real input and one measurable output. A working two-step pipeline beats seven installed repos doing nothing.
These are the official repositories. Open the one attached to your biggest bottleneck, follow its setup block below, and bookmark the rest for later.
Pro tip: Every link above goes directly to the official GitHub repository. The setup commands below are taken from each repository's current README.
by Divjot Sahni · EZYE Consulting · ezye.com.au
Start with the bottleneck
CRM first when leads disappear between conversations
Research first when decisions rely on shallow AI answers
Knowledge first when the same files get explained repeatedly
Creation first when scripts are ready but assets are slow…
Connect the seven repos
Capture demand in DeskcommCRM
Research in parallel with OpenResearch
Convert and remember context with MarkItDown and WeKnora
Write, voice, and render with Humanizer, VoiceStudio, and HyperFrames…
Free access — enter your email below