Available for new projects

The work between your tools is still being done by hand.

I build the automation that takes over that work for SaaS and e-commerce teams — onboarding, support, retention — with real error handling and documentation your team can maintain.

Send one process — free Watch four systems

Built with the tools your stack already runs on

The problem

Where the hours actually go

Not in the hard problems. In the handoffs between them.

01

Someone re-types the same data into a second tool every morning.

The record moves the moment the event fires, validated before it lands.

02

Urgent work is found by a human reading everything that arrived.

Each item is classified on arrival and routed on content, not keywords.

03

An automation fails quietly at 3am and nobody notices until Monday.

Failure paths alert a channel, log the reason, and resume where they stopped.

The work

Four systems, built end to end

Each one designed and built in-house to demonstrate an approach — not client work. Every one has a full walkthrough, so you can see exactly how it's put together before you ever reach out.

1 Customer support

Self-hosted AI support agent

Support teams drown in repeat questions, and generic bots invent answers when they don't know one. This agent answers only from the company's own policy docs and looks up order status. When it can't answer, it says so — and posts a categorised handoff to Slack with the question, the reason, and the order number. Runs entirely on your own infrastructure. No customer data leaves it.

n8nQdrantOllamaGroqSlack
Walkthrough Self-hosted support agent
2 Retention

Onboarding signals and churn risk

Most SaaS churn is decided in the first 30 days, and stays invisible until the cancellation lands. This system scores each account's churn risk from live onboarding behaviour and triggers intervention while the account is still recoverable. Behaviour-triggered, not time-based: a user who stalls on day 2 gets help on day 2 — not in next week's report. High-risk accounts surface in Slack with full context and become assigned tasks.

Make.comGroqClickUpSheetsSlack
Walkthrough Onboarding & churn risk
3 E-commerce CRM

Customer tiering and lifecycle messaging

Most e-commerce teams treat every customer the same, because their data can't tell them apart in time. The buyer who has ordered five times gets the same message as a one-time discount hunter. This system identifies who each customer is the moment the order event arrives — not overnight. Data is validated before it touches the record, customers are routed by tier, and each tier gets its own generated message.

Make.comGroqSheetsGmailSlack
Walkthrough Customer tiering
4 Support operations

Ticket triage and routing

Support teams don't drown in hard tickets — they drown in sorting. Everything arrives looking identical, so finding the urgent one means a human reads them all. This system classifies every ticket the moment it arrives and routes it on content, not keywords. Urgent and normal follow separate paths. If the AI call fails, a dedicated failure branch alerts Slack and resumes — no ticket disappears silently.

Make.comZendeskGroqSlackSheets
Walkthrough Ticket triage & routing
Free

Send me one process. I'll record how I'd automate it.

Pick the thing your team does by hand every week and describe it in a few sentences. You'll get back a screen recording — under five minutes — showing where it actually breaks and how I'd rebuild it. No call, no pitch, no obligation.

  • Your process, restated — so you know I understood it before anything else
  • The exact points where a person is copying data or waiting on someone
  • The solution sketched on a real n8n or Make canvas, not described in words
  • What it would take to build, roughly what it costs, and whether it's worth doing at all

This covers one process. If you want your whole operation mapped and prioritised, that's the workflow audit below.

Where to start

Start with a workflow audit

Manual handoffs between tools quietly cost hours every week, even when nothing looks broken. The audit shows you exactly where those hours are going, and which fix is worth doing first.

Audit From$99

I map how work actually moves between your tools, find every place a person is copying data or waiting on someone else, and score each one on time cost, how often it repeats, and how reliably it can be automated.

  • A written blueprint: what to automate, and in what order
  • Which tool fits each job, and roughly what it saves per week
  • The tasks you should leave alone — automating the wrong thing costs more than doing it by hand
  • Delivered in a few days, from a short questionnaire. No calls needed.
This is the audit, not the build. You can take the blueprint to any developer. There's no obligation to hire me for the work.
Start an audit — from $99 Book it on Upwork
How I work

How the work runs

Written specs and walkthrough videos instead of standing meetings. Calls when they help, not by default.

01

You describe the bottleneck

A short form or a message. What's manual, how often, and who does it today.

02

You get a scoped plan

A walkthrough of the approach and a fixed price, before anything is built.

03

Build and edge-case testing

Built in n8n or Make, then tested against what breaks things — bad data, failed API calls, empty responses.

04

Handover

Documentation and a walkthrough video, so your team can maintain it without me.

I'm Ali, an automation engineer working with SaaS and e-commerce teams.

Before you ask

Questions people ask first

Which platform will you use — n8n, Make or Zapier?

Whichever fits your stack and budget, not whichever I prefer. Make is fast to ship and easy for a non-technical team to read. n8n is worth it when you need self-hosting, data privacy, or you've outgrown task-based pricing. I'll say which one and why before any work starts.

Why would we move off Zapier?

Usually cost and ceilings. Task-based pricing scales badly once volume grows, and complex branching or error handling gets awkward. n8n can be self-hosted with no per-task fee. That said, if Zapier is already working for you, moving is not automatically worth it — the audit tells you either way.

What happens when an automation breaks?

Every build gets explicit failure paths: the error is logged, a channel is alerted with the context, and the run resumes or queues rather than vanishing. Silent failure is the difference between a demo and a system you can rely on.

Do we need to be on calls?

Not for the audit — that runs entirely from a written questionnaire. For project work, most of it runs on written specs, documentation and walkthrough videos, which is faster for both sides and leaves a record your team can go back to. If a call helps at the start of a build, that's fine too.

What do we get at the end?

Working scenarios in your own account, documentation explaining how each part works, and a video walkthrough. Automation that only one person understands is a liability, not an asset.

Get in touch

Tell me what you're after

Pick what you're after, then describe the process. I'll reply within 2 working days.

Work email only — that's how I'll get back to you.

You'll get a reply within two working days.

Not sure where to start?

Start with the free teardown on one process — no call, no obligation. If there's more to fix than that, the audit maps the whole operation and tells you what to do first.

Send one process — free See the workflow audit