I help businesses find where AI fits in their operations — then I build and maintain it.
[ My story ]
For three years I built AI systems for other people’s companies — architectures, agents, products. Shipped, broken, rebuilt, repeated.
Eventually I landed on one conclusion: real AI doesn’t come from the tools. It comes from the engineer behind them.
In that same time, I watched the same pattern repeat. A business gets excited about AI. They hire someone to build them something. Months later, the architecture is wrong, the use case didn’t even need AI, or the product just doesn’t do what they were promised. Time and money — gone.
So I left my job to do this full-time. To be in the room when the decisions get made. To build the right thing, not the impressive thing. To have full control over what gets shipped — and full accountability for whether it works.
Today I build AI SaaS products, AI automations, and custom agents — tuned and tested until they’re accurate enough to actually run a business on.
But the most valuable thing I bring isn’t the code. It’s knowing when AI is the answer — and when it isn’t.
One engagement. Three phases. Fixed scope, fixed price, no surprises.
I spend two weeks inside your operations — workflows, tools, bottlenecks. You get a prioritized roadmap of what AI should and shouldn’t touch, with effort and ROI estimates.
If the roadmap makes sense, I build it. Workflow automation, custom AI agents, integrations with your existing tools.
Shipping isn’t the end. I stay close after launch — making sure everything runs smoothly, fixing what breaks, and giving you a direct line when something changes in your business that the system needs to handle.
Meeting 01
We sit down together and I learn how your business actually runs. Not the org chart — the real flow. What tools you use, where your team’s hours go, which tasks repeat, where things break. By the end of this meeting I know your operations well enough to ask the right questions in the next one.
Meeting 02
Now we go granular. Workflow by workflow, I separate what’s worth automating from what isn’t. Some things look automatable but aren’t — too much human judgment, too much variance. Some things look small but quietly cost you a day a week. You’ll leave this meeting knowing which is which.
Meeting 03
I come back with a written roadmap and architecture of what I’d build for you — If something on the list doesn’t make sense to you, we kill it. If the whole thing doesn’t make sense, you keep the roadmap and we part on good terms. No pressure to move forward.
Then — the build
If you approve the scope, I build. Four weeks later, you have AI running the operations we agreed on — tested, deployed, documented, and connected to your existing tools. And I don’t disappear after launch — we stay in touch so I can make sure everything keeps running the way it should.
[ Past work ]
Phone agents that answer calls and book appointments.
Retrieval that reasons across your data.
Local Open-source models finetuned on your data.
Automations wired into the tools your team already uses.
Multi-step agents that run processes.
Production AI products end-to-end: model, backend, UI.
Tell me a bit about you and I’ll send a link for a free 20-min call within 1–2 business days.
© 2026 Salah-Eddine Khalil · Rabat, Morocco