AI automation services · integration · custom AI solutions

AI automation,
without the agency.

You searched for an AI automation agency. What you actually need is the work an agency would hand to its one good engineer: find the task your team repeats every day, wire a model into the tools you already use, and ship it so it runs without anyone babysitting it. I am that engineer, and you talk to me directly.

The work

Four kinds of automation that earn their keep.

Each is scoped to ship on its own. Start with the one that removes the most hours from the week; the others can wait until the first has paid for itself.

The receipts

Automations already running in production.

How it runs

No retainer, no discovery phase, no surprise invoice.

Written up properly

Three automations, written up in full.

The questions

Asked before every first call.

Not sure the idea deserves a build at all? The small-business page is about picking the one automation worth starting with.

Are you an AI automation agency?

No, and that is the point. An agency sells you an account manager and passes the work to whoever is free. I am one engineer: I scope it, build it, deploy it and answer your messages. You get agency-grade output without paying for the layers in between.

What can actually be automated with AI?

Work that is repeated often and tolerates review: answering recurring questions, qualifying and routing leads, drafting documents from templates, extracting data from files, searching your own knowledge, producing content variations. Work that is rare, or where one wrong answer is expensive and nobody checks it, is a bad candidate, and I will tell you so.

Do you use n8n, Make or Zapier, or do you write code?

Whichever fits. n8n is a good choice when your team wants to see and edit the workflow themselves. Code is the right choice when the logic is complex, the volume is high, or the automation is part of a product. Many builds are both: a coded core with a visual layer on top.

Which AI models do you work with?

Mostly Claude and Gemini, chosen per task: a strong model where reasoning matters, a small fast one for the high-volume steps. The system is built so the model can be swapped without a rewrite, because prices and quality change every few months.

How do you stop it from inventing things?

By engineering, not by prompt wording. Answers are grounded in your documents with the source shown, a second step checks the first, and a human approves anything that leaves under your name. On FASL, legal drafting where a wrong answer is costly, that structure is the whole product.

What does AI automation cost?

It depends on scope, so a number on a web page would be a guess. The shape is fixed: a free call, then one price agreed in writing before the build starts. Running costs (model usage, hosting) are estimated in the same document so there is no surprise later.

Will it work in French or Arabic?

Yes. I work in English, French, Arabic and Darija, and I have shipped systems in all four: FASL runs Arabic-first, RESO Khdma talks to users in Darija, Belmo sells in French and Arabic.

Who owns the automation when it is done?

You do: the code, the workflows, the prompts and the accounts. It runs on infrastructure you control. Nothing is locked to a platform you have to keep renting from me.

Free call · no deck, no pitch

Tell me which task
you would delete.

Thirty minutes, no charge, and an honest answer at the end, including “do not automate this.” I read and answer every email myself, within 24 hours.

Email [email protected] →