Find the opportunity
Four to six interviews, usually across three to four functions
When the starting point is unclear, I compare candidate workflows and recommend the one worth redesigning first.
A paid extension to the design engagement.
Buying AI is easy.Getting faster isn’t.
AI makes tasks faster. I redesign the whole workflow, using process change, automation and AI to make the business faster.
Where the work waits
Faster tasks don’t necessarily make a faster business. The bottleneck is often between the tasks.
A simple process, held up by handoffs and waiting
Four people, four tasks. Each of them is quick. Most of the time, the work is waiting between them for the next handoff.
Time from request to done: this is the first bar, the one the others compare with.
Adding AI at each step barely speeds up the whole
Give each person AI and every task gets quicker. The waits stay where they were, so the work arrives barely sooner.
Time from request to done: 1.2× faster than the simple process.
In reality it is worse: most processes are far more tangled
Several processes run into each other. A request waits for every part it depends on, then waits again at every handoff.
Time from request to done: 1.6× as long as the process with faster tasks.
Improving it significantly takes a view of the whole system
The same work and the same people, redesigned as one system. Most of the waits are gone.
Time from request to done: 2.1× faster than the system before.
Illustrative simulations, not measured data or a forecast. In every stage the same work arrives at the same pace; each handoff collects only a few pieces at a time.
Illustrative simulations, not measured data or a forecast. In every stage the same work arrives at the same pace; each handoff collects only a few pieces at a time.
So I look at the whole path, and ask three questions of every step:
- What still needs a person?
- What can the system do?
- What no longer needs to exist at all?
the one faster tasks never answer
Where I work
Answering those questions takes process improvement and AI together. I combine experience in both, with data and analytics, to address the problem as a whole.
Process improvement
Lean Six Sigma
I find where work waits, loops back or varies, and fix the cause. I am a Lean Six Sigma Black Belt.
Automation
Where a simple rule will do, I automate it before reaching for AI.
Where the two combine
Workflow redesign
I change how the work moves between tasks, not only how each task is done.
AI, data and analytics
AI and agentic systems
I build AI that reads, drafts and acts across your systems, with clear permissions and a named owner.
Data strategy and governance
I help you decide which data to invest in first, and make ownership and definitions clear.
Advanced analytics
I build models where they change a decision, such as demand, risk or which case to handle first.
What I do
Before, if needed
The core engagement
Design the solution
About 1.5 to 2 weeks
One workflow, redesigned so the work waits less between tasks, and ready for your team to put in place.
I follow the whole path of a request, across the teams and systems it touches, and design how it should run, drawing on process change, automation, AI, data and analytics as the workflow needs.
The design sets out:
- who does what
- the systems and decisions involved
- controls, escalation and ownership
- dependencies
- how success will be measured
- the order to put it in place
After, where justified
Build the AI components
Scoped separately, once the design is agreed
I build, test and document the AI parts of the design, then hand them over to your team with the training to own them.
AI can do the work. It cannot hold responsibility for it. Every build makes three things explicit: what the AI may do, what it must escalate, and who remains accountable.
Your team and existing partners carry out the other process and system changes, and I work with them on what the AI depends on. Support after handover is agreed case by case.
After, or on its own
Fractional AI leadership
Once we have worked together, or on its own from the start: typically a few days a month
Senior AI judgement for your leadership team, without a full-time hire. In effect, a fractional Chief AI Officer.
I help you prioritise opportunities, assess proposals, set data and AI governance, shape your data strategy and decide what to pursue, sequence or stop. It does not require a project first.
The person you meet is the person who does the work
I’m Laurent Tironi. L’Atelier is my independent practice, based in Dubai. I work with three clients at a time, so the work gets my full attention.
Before L’Atelier
- 18 years redesigning how organisations work, 9 of them in data and AI
- At Etihad Rail, built the Data & Analytics function
- At QNB, pioneered a retail AI use case and helped establish the case for an analytics team
- Five years of Lean consulting at Capgemini: factories, hospitals, public services
- Six years in banking, including a bank-wide automation programme
- Lean Six Sigma Black Belt
If you already have data, AI or technology teams, I work alongside them. LinkedIn profile
Start a conversation
Bring a workflow that is slowing you down, a data or AI question you have not pinned down yet, or a need for senior AI guidance. We talk for an hour about how things work today and whether I can help. No charge, no obligation.
Start a conversation
Or write to laurent@latelier.build
Prefer to read first? The brochure, or the latest note, Why AI automation stalls in growing companies.