52tools
run the average UK small business, with 27 separate logins. Even a sole trader working alone juggles 44.
Sage · The Hidden Cost of App-Switching, 2026Hi, I'm Achraf Ouahab
I build agentic AI systems that scale business operations on infrastructure you own.
When you own the infrastructure, your AI agents can work across your whole business from one database: following up on leads, planning content, chasing late projects and summing up your week. You pay hosting and services directly, with no platform fee in the middle. And before anything is sent, published or paid, they ask you first.
Founder of ACHRAF OUAHAB MEDIA LLC.
Why own your infrastructure
Own the infrastructure, and your AI agents can work across the whole business.
AI agents can now plan work, do it, check it and report back. What they can do depends on what they can reach. On rented tools, each vendor decides that. On infrastructure you own, you do.
Before
6 tools · 6 bills
Six rented tools
New report? You ask the vendor.
After
0 platform fees
One foundation you own
New report? Your agent builds it.
-
What you get
Your agents work across every part of your business: every lead, client, project and report in one database, one set of rules that you set, and code they can change after your OK.
-
What you stop paying
Rented tools often charge more for every user, subscriber or sale. On infrastructure you own, you pay hosting and services directly, at their own prices, with no platform fee in the middle.
-
What you stop losing
You stop losing hours to copying data between tools that don't connect, and no vendor decides what you can add or take with you. Your code, hosting and data are in your name.
The numbers · 2026
Renting costs time. And it holds AI back.
Four numbers from this year's research: what juggling tools costs small businesses, why scattered data holds AI back, and how fast AI agents are arriving anyway.
19days
of work a year lost to moving and re-entering data between tools that don't connect: 2.9 hours a week. Nine in ten UK small businesses lose time this way.
Sage · The Hidden Cost of App-Switching, 202642%
of UK small businesses want to use AI more but say their financial data is too spread out for it to be useful. 53% would be more confident using AI with their data in one place.
Sage · The Hidden Cost of App-Switching, 20264×
as many large US companies are building or deploying multi-agent systems, where several AI agents work together, as a quarter earlier: 25%, up from 6%.
KPMG · AI Quarterly Pulse, Q3 2026Agents are arriving fast. Scattered data holds them back. Infrastructure you own gives them one place to work.
See what the agents doThe agentic AI systems
Four systems that take on the work, under rules you set.
In plain words
An AI agent is software that takes a goal, plans the steps and does the work with your tools, instead of only answering questions. An agentic AI system gives each agent one clear job, only the access that job needs, and rules for when to ask you. Because all four run on infrastructure you own, they work from the same data, under the same rules.
-
01 · Builds and changes your infrastructure
The build agent
You describe a change in plain words: a new field on your leads, a new report, a new client dashboard. It builds and tests the change in a secure cloud workspace, shows you a private preview and puts it live after your OK. This website is built this way today.
-
02 · Runs your operations
The operations agents
A coordinator takes your request and hands the work to four specialists, each with access only to its own job. They run when you ask, on a schedule, or when something happens, like a new lead. Before anything is sent, published or paid, they ask you first.
Coordinator
- Leads and clientsqualifies new leads, drafts follow-ups, keeps your CRM up to date
- Contentplans what to publish, drafts posts and emails in your voice
- Projectstracks tasks and deadlines, chases what is late
- Reportssums up your week in plain words
-
03 · Answers your visitors and customers
The customer-facing agents
The visitor agent answers visitors from your own pages and adds each new lead to your CRM. The member agent helps customers find what they paid for, and hands over to you when a person should answer. Both answer only from your own content and never make promises for you.
-
04 · Moves your data
The flows
Not agents: fixed automation, and I say so. When a customer pays, their access opens, the client is added to your CRM and the welcome email goes out. Nothing is copied by hand, and the agents work on top of the flows.
As your leads, clients and content grow, the repeated work goes to agents instead of onto your desk, and you still approve what matters.
One rule for every agent. An agent writes down every step it takes. And before anything is sent to a customer, published on your website or paid, it stops and asks you first.
See what they run onWhat they run on
One foundation every agent works from. Yours, not rented.
This is what your agents run on, the Front-to-Back 3-Layer Infrastructure: one foundation that holds all your data, with three layers on top, one for each group of people your business serves. It is not an all-in-one platform, which rents you its features on its prices and its rules. It is infrastructure you own, built around how you work.
-
Website
Front · For your visitors
Where people find you and become leads and customers. The visitor agent answers them here, and every lead lands in your own database.
- Pages and blog
- Lead generation funnels
- Sales funnels
-
Member portal
Middle · For your customers
Where customers log in and get what they paid for, under your name. The member agent helps them find it.
- Courses and programs
- Resources and downloads
- Client dashboards
-
Admin portal
Back · For you
The command center, where you run the business, built around how you work. Your operations agents work here with you.
- CRM
- Content planning
- Project management
-
Foundation
Underneath · Owned by you
One place for all your data, so every agent and every layer works from the same records. A sale on the website opens the customer's access in the member portal and shows in your admin portal at the same moment.
- Database
- File storage
- Payments
- User accounts
Your code
In your own repository. Every change is saved and can be undone.
Your hosting
On a global network, in your name, under your own domain.
Your data
In your own database, in one place. You can export it or move it whenever you want.
Your OK on every change
An AI agent builds and tests each change in a secure cloud workspace, then shows you a private preview. Nothing goes live without your OK.
This website is built this way: an AI agent builds each change, I approve a private preview, then it goes live.
How it gets built
From rented tools to agents on infrastructure you own.
Four steps in plain words. You review private previews along the way, and nothing goes live without your OK.
- 01
Map
Find where the work piles up
How your business runs today: the tools you pay for, the handoffs between them and the work nobody owns.
- 02
Plan
Agree on what gets built
The agents you need first, the points where they ask you and the parts of the infrastructure they need. In writing, before anything is built.
- 03
Build
Build and test
An AI agent builds and tests every part in a secure cloud workspace. You review private previews along the way.
- 04
Hand over
Own it and run it
Your code, hosting, data and agents, in your name, with documentation. Then you run the business by talking to your agents.
Three ways to get it
Done for you
I set up your agents and build the infrastructure they run on, then hand it all over.
Starts with an application
Done with you
We build it together in working sessions, and you learn to run it as we go.
Starts with an application
Do it yourself
A step-by-step guide to build it yourself, with your own AI agent.
One fixed price
01Planthe steps
02Acton your infrastructure
03Checkagainst your rules
04Askfor your OK
The person behind it
Achraf Ouahab, founder.
- What I do
- I build agentic AI systems that scale business operations on infrastructure you own.
- What I believe
- A business should own how it runs. Agents do the work, people approve, and the result belongs to the business.
- What I won't do
- Put a platform between you and your business, or leave you dependent on me.
LinkedIn recommendations
What people say about working with me.
-
Bill Dickinson, D.Min., RCC
Senior Leadership AdvisorFounder, C3 Leadership
“Innovative. Strong Work Ethic. Patient. Learning Agile. Not only is he creative, but also organized. Achraf has mapped a planning process for his work with me which is impressive and confidence-building. I cannot recommend him more highly.”
-
Corinna M. Hagen
Executive Coach & MediatorFounder, ZARADIGM
“Achraf is supremely organized and able to handle complexity. He has shown to be a great listener and a strategic thinker with a high level of creative thinking skills. I would recommend Achraf to anyone looking for social media marketing support!”
-
Muktak Swami
Digital Strategist & Operations Leader, Salesera
“Achraf consistently impressed with his ability to deliver high-quality content that not only met deadlines but also resonated with our target audience. He quickly grasped our brand voice and consistently delivered content that aligned with our marketing goals.”
-
Brandon Gurley
Chief of Police, Brookhaven Police DepartmentFounder, BRG Leadership
“Achraf redesigned my website and completely nailed the vision I had for my brand. He listened, understood what I was going for, and delivered work that exceeded my expectations. If you need a website that actually reflects who you are and what you do, Achraf is the guy.”
From the blog
Notes on building with AI agents.
Plain-language notes for business owners: what AI agents can do, where they fit and why they work best on infrastructure you own.
All posts →Ownership
Rented tools or infrastructure you own?
Rented tools are fast to start, but each vendor sets the features, the fees and what your AI agents can reach. When does owning your infrastructure pay off?
Governance
Guardrails and decision rights: keeping AI agents accountable
How to keep AI agents reliable and accountable: least-privilege permissions, approval points, testing against real scenarios and a full audit trail.
Agentic AI
What makes an AI system agentic?
A chatbot answers. An automation follows a script. An agentic system takes a goal, plans the steps, acts across your systems and checks its own work.
Questions
Asked before you ask.
Something else on your mind? Email [email protected] and you'll get a straight answer.
See what I build →Because your AI agents can only work with what they can reach. On rented tools, an agent connects to each tool through the door its vendor built for software, an API or an MCP connection, on that vendor's terms: part of the data, read only, or only on a bigger plan. Your data also sits in several places, joined by syncs that can break. On infrastructure you own, your agents work from one database, under one set of rules you set, and the code, hosting and data stay in your name. You pay hosting and services directly, with no platform fee in the middle.
Four systems, all built in. The build agent builds and changes your infrastructure, and nothing goes live without your OK. The operations agents run your operations from your admin portal: a coordinator hands work to four specialists (leads and clients, content, projects, reports). The visitor agent answers visitors from your own pages, and the member agent helps customers find what they paid for. The flows are fixed automation, not agents: a sale on the website opens access in the member portal, adds the client to your CRM and sends the welcome email.
Each agent gets only the access its job needs. It writes down every step it takes, and before anything is sent to a customer, published on your website or paid, it stops and asks you first. This is called human-in-the-loop approval, and everything I build has it.
The Front-to-Back 3-Layer Infrastructure, owned by you: a website for your visitors, a member portal for your customers and an admin portal for you, the command center, on one foundation that holds your database, file storage, email, payments and user accounts. Because it is one foundation, every agent works from the same data.
A platform rents you its features: its prices, its limits, its rules, and your data inside its walls. This is infrastructure you own: the code in your own repository, the hosting in your name, the data in your own database. You can change anything, add what your business needs next, and move it whenever you want.
You own all of it: the code, the hosting and the data, in your name. Each month you pay the services it runs on directly, at their own prices: hosting, email, payment processing and your AI agent's subscription. There is no platform fee in the middle.
No. You describe what you need in plain words and approve private previews. An AI agent does the building. At handover you get written documentation in plain language and a short training on how to work with your agents.
Three ways. Done for you: I set up your agents and build the infrastructure they run on. Done with you: we build it together in working sessions. Both start with an application, and I reply in writing with a scope and a price. Do it yourself: a step-by-step guide to build it with your own AI agent, for one fixed price. The details are on the What I build page.
Start here
Let agents run your operations. Own what they run on.
Your agents follow up on your leads, plan your content, track your projects and put your week on one page. You ask, they do the work, and nothing is sent, published or paid without your OK. The code, the hosting and the data stay in your name.