AI transformation, built and run with you

Put AI into the business
you already built.

We find the one process in your company worth putting AI into first, then build it with you. It starts with an audit. The price is $2,500–$5,000, agreed with you before any work starts, and it ends on day 14 with a written plan and the sums shown. If we find nothing worth improving, you don't pay.

Companies of 20 to 200 people. Not startups, not large groups.

Apply for an audit · from $2.5K Read what broke

Arun Agrahri & Cédric Giorgi

"We don't sell what we haven't lived. Two of us run this whole company with no operations staff: the AI processes every meeting the same night, writes the Sunday report, and keeps the CRM honest. We built ours in the same order we will build yours."

The audit

First call: free, and you sign nothing
Price: $2,500–$5,000
Ends: day 14
You get: a written plan, with the sums
If nothing is worth building: you don't pay

Then the build, then we run it, both priced from the audit. Stop whenever it stops paying.

Two of these are already in your inbox. Neither one has your company in it.

AI Transformation Strategy

214 slides · phase 1 of 4 · month six

New message · someone you've never met

"Hey! I noticed you run a company. I can automate everything you do by Friday. Quick call this week?"

A third offer is coming from the company that already sells you your phones and your printers. They will get the meeting, because you already pay them. They can sell you licences and a training day. None of the three writes down how your company decides, and that is the part the AI needs most.

There is another way to start. Fourteen days, one price you can see. You come out knowing the one process to begin with, why that one, and what a year of running it the current way costs you. Then you decide.

How it works

The AI goes in last. First we write down how the work already happens.

Your company already runs on a manual nobody has written down: where things are, how they move, and what your people decide without being asked. An AI that has not been told any of that will guess, and it will sound certain. So we write it down first, then build the systems that run it with you.

The order we build in

Seven layers. The AI is the seventh.

Most AI projects start at the last layer with nothing under it. That is why the demonstration works and the real thing does not. A tool that answers your customers is only as good as the information behind it, and the judgment behind that.

We work down the list in order. Every layer is worth something on its own, so you can stop at any one of them and keep what you have. The mark against each one shows the usual split between what the audit settles, what we build, and what we keep running. Your own order can move that line.

01

Where your information lives

One honest picture of what you know, and where it actually sits today.

audit
02

Getting your systems talking

Your existing tools, connected, so nobody retypes the same thing twice.

build
03

How your company decides

Your house rules, written down: the judgment your team applies without thinking. This is the layer nobody else installs, and it is why their answers could be any company's.

build
04

The first workflow

One process, running end to end, on the work you already do every week.

build
05

Weekly quality checks

Someone measures whether it is still right, every week, and shows you.

run
06

Your own skill library

Each thing we build stays yours, reusable, and adds to what came before.

run
07

The AI that runs it

Last, and only once the six above it are in place. It works in the background, and your team approves the results.

run

How we choose

What makes a good first one.

how much time it eats: hours a week, every week
where the information lives: your systems, or people's heads
how many people touch it: one, or a chain of them
what we come back with: the one to build first

Run those three against a process in your head right now. Hours every week, information in more than one place, more than one person touching it. If a process has all three, you have just found the one we would start with. The audit puts a number on what one more year of running it the current way costs you.

Start smaller than you'd expect.

Every engagement is fixed in scope and credits toward the next step. You are never buying a retainer on faith.

What happens after you apply

Four conversations, and the work in between.

Thirty minutes, free. Either one of us names a process worth fixing, or one of us says there is not one here and we stop. It costs you half an hour and you sign nothing.

The deep dive is where the value is: 90 minutes inside how the work actually moves through your company. You'll know the price before anyone asks you to decide, and what comes back is yours whether or not you build with us.

01 Discovery: 30 min · free · day 1
02 Deep dive: 90 min · days 4–8
03 Checkpoint: 15 min · optional
04 Walkthrough: 60 min · day 14

Audit

$2.5–5K

Fourteen days at most, one fixed deliverable. Which single process to start with, what a year of the current way costs you, and who inside the company has to own it for it to hold.

Where your operations stand today, and what waiting costs you
The order we build in: the seven layers, in the sequence that works
Fee credits 50% against implementation if signed within 30 days

$2.5K covers most companies. It moves toward $5K when more people are involved, more processes are in scope, or your information sits in more places.

What the audit costs, day by day

If it finds nothing worth building, you don't pay.

Apply for an audit

What comes after.

Neither of these is a decision you make today. You see the build price at the walkthrough on day 14, before you decide anything, and the audit is the whole commitment until then.

Implementation

scoped

4 to 12 weeks, fixed scope. We build exactly what the audit scoped, in the order it laid out, with a working demo in a shared channel every working day.

Priced from what the audit finds. You see the number before you decide anything.

Operated

scoped

The system, run for you: weekly quality checks, fixes when something drifts, and a new capability every month, so it compounds instead of decays.

Priced once we know what we are running for you. Monthly, and you can stop.

See it worked through

What the audit found in a company like yours, and what got built first.

Five worked examples, one per sector: what the audit found, what we would build first, and what the week looks like afterwards. The companies are examples built from firms like yours, for demonstration only, and the arithmetic is printed beside every figure.

Distribution
Manufacturing
Professional services
Construction
Home services

See what it found

Proof

We don't sell what we haven't lived.

Our company runs Sena, AI that matches people at live industry events, used by teams who don't write software. And the company itself runs on a private skill library: 20-plus AI systems that do the operational work of a staff we never hired.

Two founders, no operations team. The seven layers do that work.

We publish how each one works, what it cost to build, and which ones we switched off because they were not worth the attention.

// how two people run a whole company

meeting-review · every transcript processed, tasks filed. 18:00, Mon–Sat

weekly-report · Sunday synthesis, catches what we forgot

prospect-research · pre-meeting brief from a website URL

event-report · post-event analysis and CRM export

sponsor-enrichment · every sponsor researched before we ask

whatsapp-broadcast · event comms drafted and delivered

event-expert · our own analyst, briefed on the whole industry

+ the house library · research, content, SEO, outreach

your company gets its own →

Owners talking to owners.

Arun Agrahri

Managing Partner

Serial founder based in San Francisco, with two exits (Snap and Manulife). Builds the systems, writes the letters, and runs the ones behind our own company as well as yours.

Cédric Giorgi

Managing Partner

Twenty years in European tech, from Sigfox to co-founding La French Tech Toulouse. Works alongside your team on the ground, and hosts Conversations: owners of traditional businesses on what AI actually changed.

The letter

How a traditional business becomes AI driven, with proof.

Essays from running our own company on AI, and from the work inside other people's. What worked, what broke, what it cost. Written for owners, not engineers.

Conversations · the podcast

Owners like you, already running on AI.

Cédric sits down with owners of traditional businesses who put AI to work inside their companies. No vendors, no futurists, no demos: just what actually changed.

First conversations are being recorded now.

Fair questions.

We're not a tech company. Will this work for us?

That's exactly who this is for. We build around the work you already do: invoices, quotes, follow-ups, reporting, hiring. Your team keeps working the way they know; the system adapts to them, not the other way around.

Our IT provider says they can do this. Why you?

They can sell you licences and a training day, and if that's what you need, take it. Ours is a different job: we go inside one process, change how it runs end to end, and stay until it holds. Two questions worth asking them: who writes down your team's judgment so the AI applies it, and who checks every week that it's still right.

Does my team have to learn to use AI?

No. It works in the background, inside the tools your team already opens. What reaches a person is a draft, a suggestion or an exception, and their job is to approve it, correct it, or say no. Nothing new to log into, no course to sit through. Every new step starts with a person checking each result, and only runs on its own once it has been right for two weeks straight.

Why pay for an audit when free ones exist?

Plenty of firms will audit you for nothing, and three of the twelve we reviewed in August 2026 do. If a free review is all you need, take it. The difference is what you keep at the end. A free audit ends in a proposal to sell you the next thing. This one ends in a priced written plan you own, with the sums under it, and it costs you nothing if there is nothing in it worth building.

Why not a big consultancy, or a cheaper freelancer?

A consultancy will sell you a strategy before anything works. A freelancer will sell you a script without a system around it. We sell staged, fixed-scope steps where each one credits toward the next, and you can stop at any point and keep everything we've delivered.

What is the audit, and what is it not?

It is not an inspection. Nobody arrives to collect your documents and grade you. It is two conversations: 30 minutes, free, to see whether there is anything here worth doing, then 90 minutes going deep on how the work actually moves through your company. What comes back is the one thing to build first, why that one, and what it returns. Where a number can't be measured, we leave it blank instead of estimating it.

What happens when it breaks or drifts?

AI systems decay without attention, which is why the operated tier exists: weekly quality checks, fixes when something breaks, and a new capability shipped every month. If you'd rather run it yourself, you own everything we build, including the skill library.

Who owns what you build?

You do. The system, the documentation, and the private skill library: your team's judgment, written down so the AI runs things your way. We hand off cleanly whether you keep us on or not.

Two weeks. A plan you can defend.

From $2.5K, fixed scope, half of it off the build if you go ahead within 30 days, and the first call is free. You'll know what to build, in what order, and what it gives back, before you commit to building it.

Tell us the one process that costs you most. One of us reads every application and answers it with a name.

We invoice in US dollars from Sena Technologies, Inc.; the French page shows the same amounts in euros.

We reply within two business days, always with a name. No sequence, no chasing.

Put AI into the business you already built.