Expertise

AI Automation Systems

Self hosted n8n and Supabase systems that quietly run the parts of a business no one wants to touch again.

Automation you can actually own

Most no code automation ends in vendor lock and per task pricing that punishes you for growing. I build systems on self hosted n8n, deployed with Docker Compose on DigitalOcean, backed by Supabase. You keep the workflows, the data, and the runtime.

The result is infrastructure a small team can operate without a platform team. Version controlled workflows, observable runs, and predictable costs.

The parts that break in production

Real automation is not a single trigger and a happy path. It is FTP feeds that arrive late, webhooks that retry twice, records that arrive half formed, and downstream systems that rate limit at the worst possible moment. I design with those failure modes as the default assumption.

Multi branch workflows, cron scheduled reconciliation jobs, idempotent upserts into Supabase, and structured error routing are the difference between an automation that saves hours and an automation that quietly loses data.

Where I plug in

Some teams need an automation architect for a defined build. Others need a second set of eyes on a system that has grown past its original design. Both are fair engagements. If you are looking for an automation consultant who has actually operated the tools at scale, that is where I live.

Frequently asked
n8n vs Make vs Zapier, which should I use?+

n8n if you want ownership, code friendly steps, and predictable cost at scale. Make if you need a strong visual builder with rich integrations. Zapier for the widest integration surface when speed to first automation matters more than long term cost.

Why self host n8n instead of using n8n Cloud?+

Self hosting removes per execution pricing, keeps sensitive data on your infrastructure, and lets you version workflows in git. For anything past hobby volume it is usually cheaper and safer.

How do you handle failed runs in production?+

Structured error routing at the workflow level, idempotent database writes so retries do not double post, and a monitoring channel that surfaces failed runs to a human within minutes rather than days.

Can automation replace an internal ops team?+

No, and it should not try. It reliably removes 25 to 40 percent of repetitive work so the ops team spends time on judgement, not data entry.

Estimate the impact

Move the sliders. See the shape of the numbers.

A first pass calculator for this practice area. Useful for the first budget conversation, not a quote.

Estimate the impact

Rough what an automation could save you every month.

Move the sliders. See a first pass on hours reclaimed and the true monthly cost of running it on infrastructure you own.

3,000 runs
10020,000
6 min
130
35 £ / hour
1590
// results, recomputed live
  • Hours reclaimed per month300 h
  • Labour value returned£10,500
  • Self hosted infra runtime£93 / mo
    n8n on a DigitalOcean droplet plus Supabase
  • Net monthly upside£10,407

First pass numbers, not a quote. Real builds also carry a fixed design and implementation cost that pays back inside the first quarter for most teams.

Other pillars

Continue reading