A growing library of working solutions for the operational pain points we run into most often when building automation for SMBs. Each recipe is a real problem, a plain-English explanation, working code, and the common gotchas we've hit. MIT-licensed. Use it, fork it, ship it.
This repo exists because the same problems keep showing up in small and mid-size businesses and most "solutions" online are either vendor pitches or theoretical write-ups. Here you get the actual code we use with clients, published so any in-house dev, ops lead, or founder can self-serve.
Avalux — a full-stack AI automation agency based in Mississauga, Ontario, working with SMBs in freight, e-commerce, home services, and adjacent verticals. We publish recipes from real client work (anonymized) and from our own ongoing R&D.
If a recipe almost-but-not-quite solves your problem, you have two options: fork it and adapt it yourself, or hire us to build the production version. Both are valid.
- tech-on-the-way-text — Auto-text your customer with the tech's name, photo, and live ETA when a job is dispatched. Companion to most field-service CRMs (ServiceTitan, Jobber, Housecall Pro) via webhook + Twilio.
- freight-eta-customer-text — Geofence pickup/drop, auto-text receiver when truck is one hour out. Works on top of Geotab, Samsara, or Motive APIs. Companion to our freight-eta-toolkit.
- shopify-qbo-line-item-sync — Idempotent line-item sync from Shopify orders to QuickBooks Online with proper account mapping. Companion to our shopify-quickbooks-sync toolkit.
- google-review-low-rating-alert — Get a Slack ping the moment a new Google review under 4 stars lands, with the review text and a link to respond. Works across multiple locations.
- lead-router-by-postal-code — Inbound webhook → postal code lookup → route to the right rep's Slack/email with full context. Eliminates manual triage.
Each recipe directory has:
README.md— the problem in plain English, who has it, when you should and shouldn't use this approach, common gotchas- Working code (
script.py,workflow.json, orimplementation.md) you can copy directly - A "where to put this in production" note
Most recipes are designed to be deployed on:
- A small VPS (Hetzner, Vultr, DigitalOcean — $5-10/month)
- A self-hosted n8n instance
- Or wired directly into your existing stack via webhooks
- Not a product. No subscription, no signup, no telemetry. Clone it, fork it, do whatever.
- Not generic AI slop. Every recipe is a real pattern we've shipped in production. No "AI receptionist for HVAC" boilerplate.
- Not closed source. MIT-licensed. We make money by building the production version for clients who don't want to self-host.
Run into a problem one of these recipes solves but find a bug or want to add a variation? Open an issue or a PR. We review them.
- freight-eta-toolkit — Geotab / Samsara / Motive utilities for SMB freight brokers
- shopify-quickbooks-sync — Shopify ↔ QBO middleware
- n8n-self-hosted-toolkit — Production n8n setup with hardened defaults
- avalux-open-source — Hub for everything we publish
MIT. Use it, ship it, sell it. Attribution appreciated but not required.
If you'd rather have any of these built, deployed, and operated for you — that's our day job. Fixed-scope contracts, real APIs only, monthly retainers for ongoing operations. Pricing starts at $5,000. See avalux.io or email eli@avalux.io.