Digital right hand
Your site, your funnels, your payments, your tools, your automation. I take the lot and I keep it running. What you get back is measured in days, every month.
For independents, founders and owners of companies of up to fifty people, who decide alone and have no IT department behind them.
The principle
The difference fits in one sentence: a supplier delivers and leaves, a right hand stays and looks after it.
My name is Mathéo Vincent. I am independent and I work with a small number of owners at a time, which lets me know their business almost as well as they do.
I do the two jobs people usually separate: the one who designs, and the one who runs it afterwards. What I like best is the moment a task somebody used to do every Monday disappears for good.
You no longer have to coordinate a developer, an agency, an integrator and three subscriptions that ignore each other. You write to me, and it is handled. I know your business, your tools and your habits, so nobody has to explain the context again each time.
I work with owners who have neither the time nor the appetite to become technical, and who simply want it to work, all the time, without thinking about it.
If you already have a large CRM and a team to keep it, you do not need me. You need an integrator. I work with the ones who grew by stacking tools that worked perfectly well back when they were on their own.
In practice I take responsibility for your online presence in the broad sense: what your clients see, what they fill in, what they pay, what they receive afterwards, and everything running behind the scenes to make that happen.
And I stay after launch. That is where most projects die, and it is precisely the part I sell.
What I take on
You switch on what you need. Most of my clients start with one, then hand me the rest over the following months.
Brand site, sales pages, learning spaces. Written, designed, put online, then kept up to date. With the technical search work needed to exist on search engines, and now inside answer engines too.
From the first click to the payment: short forms, qualification, instalments, and access delivered within a second of the payment clearing. Walked end to end on a phone before it opens to the public.
Your calendar ignores your invoicing, which ignores your mailing tool. We connect them. No more double entry, no more file retyped on a Friday evening, no more figure that changes depending on the screen you look at.
Assistants that read your mail, sort it, answer the routine questions, chase at the right moment and write in your voice. With one simple rule: nothing irreversible without your written go-ahead.
One screen that says where the money comes from, what will be missing in two months, and which client is drifting before they are gone. Without opening three tools or rebuilding a spreadsheet every Monday.
Servers, domain names, certificates, verified backups, monitoring. The part nobody sells because it does not show, and the part that decides whether your system is still standing in two years.
The engineering
This part is of no use to you if you do not want to read it. It is here for the moment somebody technical asks you whether I know what I am doing.
A badly designed agent does not get it wrong once. It gets it wrong six hundred times during the night, on your clients, and nobody notices until the next morning.
An agent that answers in free text is a demonstration. An agent that answers inside a schema the code can validate is a system. Everything I deliver produces a constrained output, checked before it is written anywhere, and replayed if it does not pass.
tool schema
typed fields
replay on a failed check
No irreversible action without an explicit written confirmation. The model proposes field by field, a person accepts one at a time, and what was done stays readable months later. That rule of refusal is what makes an agent deployable inside a company with something to lose.
refusal rule
field-by-field approval
readable audit log
I have built where the data policy forbids off-the-shelf connectors, where the sandbox does not talk to production, and where every boundary crossing is controlled. Development there runs end to end on fictitious data: nothing real ever leaves the client, and that is a design constraint rather than a temporary precaution.
segregation
fictitious data in development
documentation for IT review
An agent running without supervision is not an agent, it is a bet. Health probes, logs a human can read, cost per run, and an alert when a source goes quiet. I also watch what nobody watches: certificate expiry, the availability of every published address, and the staleness of figures written into your content, which quietly become false over the months.
health probes
cost per run
alert on a silent source
An event replayed twice must not bill twice. A refund must not stay counted as a sale, when it lives neither on the invoice nor on the payment session. A cancellation must close the entitlement on time, with no forgotten grace period. These are the details that separate a system that holds from one that only looks like it does.
idempotency
replay without side effects
continuous reconciliation
Nine health authorities followed every day, from the FDA to the European Medicines Agency, each with its own protocol. Collection, deduplication, prefilter, then real impact scoring by a model. A verification step tells a network outage apart from a reasoned rejection, so an unreachable source is never quietly filed as irrelevant: that is the part everyone forgets, and the one that decides whether the system can be trusted.
multi-protocol collection
impact scored by a model
an outage told apart from a rejection
The most demanding sector I work in: the pharmaceutical industry, where regulatory constraint and data governance settle the architecture before the first line of code.
The time you get back
It sits in the gestures you and your team repeat every week without ever counting them. Here are the ones I find everywhere.
Add those six lines up over a month. That is exactly what I give you back.
How it works
Nothing to prepare. You tell me about your week and your tools, I tell you what can disappear and what cannot. If it is not my trade, I say so and point you elsewhere.
I record what is connected and what is not at your end, and I estimate what each gap costs you over a year. You keep the document, whether we work together or not.
We start with what costs you most, and it goes live before we move on. Each stage ends with a live demonstration, never with a report.
A price per engagementThere is no rate card, because no two systems are alike. You get the figure after the first call, once I know what there is to do.
What stays with you
It is the first question a serious owner asks, and the only one an independent should answer before it is asked.
Repository on your account, servers in your name, keys in your hands, from day one and not on delivery. At no point does your business depend on infrastructure you do not control.
Documentation delivered as we go: what each part does, how to restart it, where to look when something is off. Written for someone who is not me.
No automation writes to your clients, deletes data or files anything without an explicit human validation. Everything that happens is logged and readable.
I am a person, not an infrastructure. I do not sell round-the-clock cover, and if someone sells you that on their own, ask them who picks up at three in the morning. What I do guarantee is that nothing I build becomes unusable if I stop.
Let us talk
Every message is read by me, never by an assistant. Answer within one working day.
Just describe the task that annoys you most in your week. It is almost always the right place to start.