How we work
From the first step to running it with you.
Every engagement runs the same four steps. It starts with Assess & Design and continues after launch, with one accountable team.
Our responsibility continues after launch.
We engineer AI around your operation and remain accountable for the systems we run, with clear controls, ongoing monitoring, and continuous improvement.
AI is easy to try and hard to trust with real work. We get you to the second part: built for your industry, guarded, and still ours to answer for after launch.
Step 1
Assess & Design
Every engagement starts here: we understand your workflows and systems, then agree what to build, where a person approves, and how success will be measured.
Step 2
Build & Integrate
We develop the AI, connect it to the platforms you already run, and build any missing systems the workflow needs.
- WhatsApp and Instagram
- ERPs
- CRMs
- E-commerce
- Spreadsheets
- Databases
- Arabic documents
Step 3
Validate & Launch
We test real scenarios with your team, verify the safeguards, and launch with agreed approval and escalation points.
Step 4
Operate & Improve
Under a yearly agreement we monitor performance, resolve issues, maintain integrations, introduce tested improvements, and report to you every month.
A report every month
One accountable team
Clear ownership from the first assessment through ongoing operation.
The first step: Assess & Design.
What the audit is.
One process
One piece of work, chosen with you. Not the whole organization at once.
With your team
We map it with the people who do the work every day, and with the person who owns it.
Before anything is built
No agent exists yet. The audit shows whether AI agents fit this work, and where.
What we look at.
Four things, in this order, because each one depends on the one before.
- 1
The process, step by step
How the work arrives, who handles it, what they check, and where it waits.
- 2
The systems and data it touches
Where the records live, how people reach them today, and what form the documents arrive in.
- 3
The rules that apply
Your own policies, plus the sector rules, contracts and data laws that limit where data may go and who may decide.
- 4
Where a person must decide
The decisions an agent may prepare but must never make alone, and who on your team makes each one.
What you get.
A written map of the process. We confirm its exact contents with you before the audit starts.
- Where AI agents can help, step by step, what each one would do, and where they shouldn't be used.
- Where a person decides, and which role on your team decides.
- The delivery option that fits your rules: your own servers, your own cloud account, or a contracted model provider.
- A pilot plan: one agent, its approval points, and how we would measure the result.
Where would you start?
Answer three short questions to see a suggested starting point. It's a suggestion, not a quote.
A good place to start
Answer the three questions to see a suggested starting point.
How long the audit takes and what it costs are agreed for each engagement.
Questions about the audit.
How long does it take?
It depends on the process. A short process on one system takes less time than one that crosses several teams. We agree the time with you before we start.
What does it cost?
It is agreed for each engagement, based on the size of the process. You see the price in writing before you commit.
What do you need from us?
Time with the people who do the work and with the person who owns it, a walk through the systems involved, and examples of the work, real or anonymised.
Do you need access to our systems or data?
Only what you agree to share. Most of a process can be mapped from conversations, the screens your team shows us, and anonymised examples.
Do we have to continue after the audit?
No. The audit ends with the written map. Whether to run a pilot is your decision.
Which process should we start with?
Usually one that repeats often, follows clear rules and slows your team down. The self-check above suggests where to look, and we choose together.
Support after launch.
What every client gets.
A named contact
One person on our team who knows your setup and answers for it.
Monitoring
We watch the agents in production and act on what we see.
The log
Every action, with its source, readable by your team and exportable.
The off switch
You can switch any agent off at any time.
Response times and availability are agreed in each contract.
How an incident is handled.
The more serious it is, the calmer and more factual we get.
- 1
Stop
We pause the agent or the step, so nothing else goes wrong. Your team can continue by hand.
- 2
Tell
Your named owner hears from us: what happened, what it affected, and what we've done so far.
- 3
Explain
From the log: what the agent did, when, and from which source.
- 4
Update
At the times we give you, until it's closed.
- 5
Review
A short written review of the cause, the fix and how we prevent it.
What we report.
- What we measured
- A note at the interval agreed in your contract: the work the agents did, the cases they handed to a person, and any errors.
- Incidents
- Every incident in writing, with its review.
- Changes
- What we changed in the agents or their rules, and why.
Questions about support.
What are your response times?
They are set in each contract, based on how critical the work is and where it runs.
Who do we contact when something looks wrong?
Your named contact. You can also switch the agent off yourself at any time.
Can we change the approval points later?
Yes. You can tighten them at any time. An agent does more on its own only when the log shows it is safe and you agree.
Talk to us about your first step.
We start by mapping one process with your team.
Talk to usYou decide. We answer for it.