Building + owning a
modern GTM OS.
How real companies rewire go-to-market around AI. One operating system, a live demo of the whole thing, the pod that runs it all.
Connect while we talk. I answer.
I build these engines for a living.
This is what I build. One GTM brain.
Built inside real companies. Two I can name.
JoomPro
Global e-commerce group. The GTM x AI core team runs scoring, CRM automation + a performance engine, weekly sprints.
Backbase
€2.5B banking software. A full end-to-end GTM engine, live: 250+ data tables, 160+ people in it daily across marketing, sales + CS.
Everhome Brands
A global consumer-goods group. Dozens of household brands, sold through every major retailer. GTM engines on the brand + retail motion.
Hotelia Grand
A hospitality-tech unicorn. A GTM engine on the pipeline + partner motion.
Cyberex AI
A cybersecurity unicorn. An OpsOS proof, same architecture pointed at operations.
Shopword
A leading European e-commerce platform. Marketing automation inside a locked-down enterprise cloud.
Voyagia Group
A €1B travel group. An SEO x GEO engine plus a voice-AI sales feasibility.
Seven bold claims for today.
The ChatGPT of the 1890s.
Product went AI-native. The rest is catching up.
At fast software companies most new code is now written with AI. Meanwhile marketing, sales + ops still run the old playbook across 30 disconnected tools. Every quarter the distance grows.
AI ROI starts in go-to-market.
Where does the ROI start? My claim is go-to-market, for two honest reasons. And there is a pattern for what happens next.
Safe to automate first
Top-of-funnel runs on low-sensitivity data. A wrong draft costs nothing before a human reads it. Legal says yes here first.
The wins are visible
Pipeline, cost per customer, output per person. Numbers nobody argues with, and visibility is what pulls the rest of the org in.
Then company-wide
I have started in ops and in finance too. Usually GTM goes first, and around month three the CEO calls about the same approach everywhere.
Real screenshots. Four engines, one curve.
Straight from Search Console. Two engines are months old, starting from zero. Two are further along. Flat while you build, then ignition, then it compounds.
One shared brain. Apps on top.
Read it left to right. Your data + your know-how feed one shared brain. The brain powers the GTM apps. A human gate sits before anything ships. Hover any block, it explains itself.
Before you ask. Yes.
Everything you are about to see was built with all of this taken into account. The full list has its own slide near the end, and we can go one by one in Q&A.
This is what it looks like daily.
A full GTM OS, live and clickable. The sum of everything running across my current deployments, fully anonymized, every number invented. We tour the engines one by one, Score, Ads, Crazy Ideas, Mission Control, SEO + GEO, Sales, then Reporting, Performance, Chat and what the whole thing costs to run.
What to expect, 6 to 12 months in.
From real projects. The banking-software numbers are public, the B2C set is an anonymized composite, and two are live right now and marked as such.
Why the shared brain changes the economics.
One version of the truth
Every app means the same thing by "account", "play" or "stage". The five conflicting spreadsheets die quietly.
Legal + IT say yes
Each app + each person sees only what they should. Governed, auditable. That is how you get a yes from legal instead of a six-week review.
~80% reuse
Each new capability reuses most of what exists. The first build is expensive, the tenth is nearly free.
Who runs it. Three seats + a coach.
Uber runs the same play. Pods, every function.
Two ways to build this. One shared ingredient.
The bulldozer in the playground
A separate environment, stood up in days. Compliant by design, signed off before a single record flows, read-only mirrors only. Speed lives here. Built for SMBs, aggressive startups + scale-ups that want proof in weeks.
Inside your own walls
The precautious road. We build inside your existing cloud, Azure, GCP or AWS, behind your access controls, at your security team's pace. Built for enterprises + regulated industries. Slower, equally real.
Build beats buy now. The honest split.
Across our current projects roughly 70% of the system is built, 30% is bought. What changed is WHERE the buying happens. You buy access to data and infrastructure. The middle, the brain and the apps, you build and own.
The thing that thinks
The brain, the apps, the agents, the model middleware. Your edge, on your data, handed over. Zero vendor lock-in, zero waiting on someone else’s roadmap.
Data + rails
Scraping, channel access, enrichment, voice, deliverability, rank data, models. You buy these because they own data or infrastructure you can’t rebuild.
Lock-in died
One unicorn client sunset €350K of annual SaaS licenses this year. Per-seat pricing is fading, thin-UI SaaS is fading. SaaS with proprietary data survives, the rest becomes an API you call.
What actually decides success.
From every project we shipped. The tech is rarely the reason these fail or fly. These eight are.
A clear mandate
A named leader who wants this and says so out loud. Air cover for when the politics show up, because they will.
Liberty to experiment
A sandbox and permission to be wrong cheaply. Teams that need sign-off to try anything ship nothing.
An AI-native GTM engineer
Embedded, on the team, stays. The scarcest resource and the single biggest predictor of success.
The right pod
Operator + ops + engineer in ONE business unit. A committee across five departments is where this dies.
Ruthless prioritization
Impact times ease, one pilot at a time. The graveyard is full of teams that started six things.
Domain expertise up front
The people who know where the money leaks pick the first plays. Engineers alone pick interesting problems, operators pick profitable ones.
Data honesty
Fix conversion tracking before believing any number. Most companies run on fiction and don't know it.
A weekly shipping rhythm
Demos every single week. Momentum IS the change management, nothing converts skeptics like shipped work.
Three moves before lunch.
Audit your data
Can you trace spend to closed revenue? Do your teams mean the same thing by "lead"? The audit takes a morning and decides everything after it.
Rank your motions
Which motion runs first, scoring, ads, outreach, SEO, sales? Impact times ease picks it. One motion, then the next. Ignore the meme generator.
Find your engineer
Inside or outside, but embedded. This is the bottleneck, so start the search Monday. Everything else on tonight's list waits for this one.
“I'm finally getting my job back and focusing on what's important.”
A real operator on a live deployment, a few weeks in. That is the deliverable. The 100x GTM operator, one person with the output of a hundred, judgment intact.
Pick your next step.
Implement this together
You want the engine inside your company. Reach out, we scope it, we build it with your team, you own it.
heyarnoux@gmail.com
Embed a GTM engineer
The scarcest hire, placed inside your team. AI-native, GTM-trained, embedded, and they stay. Also me, reach out.
linkedin.com/in/davidarnoux
Join GenAI Circle
350+ operators and founders executing this themselves. That is where we share all the playbooks, every week.
thegenaicircle.com
Build the engine.
Keep the engine.
linkedin.com/in/davidarnoux
With thanks to Oleg, Stefan and the Extrovert team for hosting. And to AI Summit Barcelona, September 22 + 23, where I teach the full course.
