About

Founder-led, and deliberately small.

VisionArchitech is an agentic AI product studio founded by Aziz Ahmed. We build AI products end to end (the agents, the models they run on, and the software around both) from London, Riyadh and Jeddah.

Founder
Aziz Ahmed
Offices
London · Riyadh · Jeddah

The founder

From the founder.

Aziz Ahmed, founder of VisionArchitech
Aziz Ahmed, Founder

I started VisionArchitech to build AI products the way I wanted them built: agents that survive contact with production, models that can run on a client's own hardware when the data can't leave it, and real applications around both, not demos that fall over the week after the pitch.

Small is a choice, not a stage we are trying to grow out of. You talk to the person accountable for the architecture, and you see software running early enough to change your mind about it.

Aziz Ahmed, founder

Where we work

Why three cities.

London is the base: UK and European work, and the time zone most of our written work happens in. Riyadh and Jeddah exist because Gulf clients ask for the same things UK clients do: someone in their own time zone, able to be in the room for a kickoff, working the same week they do, which in Saudi Arabia runs Sunday to Thursday.

It is one studio in three cities rather than three offices with three agendas. The same people scope the work, build it and hand it over, wherever you sit. Most of the delivery happens remotely; the cities matter for the parts that do not work over video: kickoffs, data and access reviews, handover.

  • London, United Kingdom
  • Riyadh, Saudi Arabia
  • Jeddah, Saudi Arabia

How we operate

Principles, stated as things we actually do.

Three of them. Each is a decision you can hold us to on a call, not a word on a wall.

01

We deploy locally by default when data can't leave.

Open-weight models (Llama, Qwen, Kimi, DeepSeek) running inside your VPC, on-prem or air-gapped. That means model selection and evaluation against your data, a deployment architecture, guardrails and monitoring, and a handover your own engineers can operate without us. When a hosted API is honestly the better answer, we say so and explain why.

02

We ship working software before we ship a slide.

The first week ends with a working spike: one real slice of the system, running against your constraints, demoed rather than described. You judge the direction from something that runs.

03

We say when AI isn't the right tool.

Plenty of workflows need a scheduled script, a database query or a rules engine. If yours is one of them, we'll tell you that, even when it means a smaller engagement, or none at all.

Honest scale

We are a small studio.

That shapes what working with us is like, in both directions. Here is the trade, stated plainly.

What it buys you

  • Direct access to the founder. The person who scopes the work stays on it.
  • No account-management layer, and no handover to a delivery team you have never met.
  • A small number of projects at a time, so the schedule we give you is one we can hold.
  • Architecture decisions made by the people doing the build, not relayed through a proposal team.

What it does not

  • No published case studies yet. The process page is what we publish instead: the phases, and the first week day by day.
  • Not a staffing firm. We take work end to end rather than by the seat.
  • We cannot take every project. If the fit or the timing is wrong, you hear it at the scoping call, not three weeks in.
  • No round-the-clock support desk. Ongoing support is an agreed retainer with named hours, or it is not promised.

One thing you can check without talking to us: the chat in the corner of this page is our own agent, running on this site. Poke it. The rest of the evidence is the process, written down.

Next

Your technology vision, built.

A short brief or a scoping call: either reaches a person, and either can end with us telling you that you do not need an agent for this.