AI and the labour market
What the forecasts say, against what the data shows.
- QuestionDoes task-level automation actually turn into growth, and for whom?
- WithProf. Alexandra Roulet, economist at INSEAD.
- StatusPublishing 15 October 2026.
Career
Business by training, technical enough to sell to the CTO.
A business degree in Montréal, then consulting and L'Oréal. A master's in computer science in Madrid. Then MuleSoft, where I was handed all 54 countries of Africa with no customers and left with customers in 14. In 2026 I left for INSEAD, to step back and build.
Chapter 1 · Business · 2013 to 2018
First, the business side.
-
2013 to 2017
Bachelor of Business AdministrationHEC Montréal · exchange at Tec de Monterrey
Co-founded a case competition with Cirque du Soleil, voted best student corporate event of the year. Won two others, and raced with the sailing team.
-
2017
Consulting internshipMonitor Deloitte
Six months as a junior digital consultant. A first look at how large companies buy technology.
-
2017 to 2018
Business analystL'Oréal · Montréal
Launched a BI platform for 100+ users across a $120M division. Enough time with the data to want to build the tools myself.
Chapter 2 · Computer science · 2018 to 2019
Then I learned how it's built.
-
2018 to 2019
Master in Computer Science and Business TechnologyIE · Madrid
Arrived on a scholarship without a technical background. Came back two years later to teach on the MBA.
-
2021
Web development bootcampLe Wagon · London
Kept going while at MuleSoft: 360 hours over 25 weeks, alongside the day job. It is why I build my own tools today.
Chapter 3 · MuleSoft, a Salesforce company · 2019 to 2026
Nobody sold here.
- 2017MuleSoft goes publicAn integration start-up lists in New York.
- 2018Salesforce buys itFor about $6.5 billion, its largest deal at the time.
- 2019I join, mid-integrationA start-up being folded into a 50,000-person company. I felt every seam of it.
Tap a country, or London
The territory
14 countries
Every shaded market has at least one paying customer.
Pick a country to follow the trip from London and see what was sold there.
London, covering Africa
Africa was a no-sell zone.
I joined while MuleSoft was still being absorbed into Salesforce, chose to stay with the MuleSoft business unit, and was handed Africa. No accounts, no playbook, no local team. I was 25, covering all 54 countries of a continent from London, and fairly sure someone would work out I was improvising.
By the time I left, it was close to $10M in annual recurring revenue.
Enterprise only: sales cycles of six to eighteen months, a few large deals a year, not volume. On each one I orchestrated a cross-functional team: solution engineers, business development reps, partner managers and value services consultants.
- 0customers on day one
- 14countries with customers
- 7-figureannual contract, the largest, with a leading Angolan bank
From booking meetings to running the market.
I joined to open doors for other people's deals. By the end I was the only MuleSoft seller covering Africa, which meant doing the jobs nobody else was there to do.
- 297%of quota in FY25, about three times my annual sales target
- Top 0.2%of roughly 25,000 sellers, selected for Chairman's Club FY25
Climb the ladder
Opening doors.
Booking the first conversations across EMEA, then leading account development for the Middle East and Africa.
3
promotions in about two years
Arthur's work in developing our business across Africa, particularly with the MuleSoft initiative, has been remarkable.
Alongside · Angel investing · since 2022
The other side of the table.
Backing founders, then helping them sell.
I started this as a side job while at MuleSoft, to learn how founders think. I invest alongside other operators through a dedicated vehicle. I bring more go-to-market help than money.
- 5early-stage startups backed
- Impact SaaS for fashion
Waro - ESG compliance
SustainSoft - Also backed3 SaaS and fintech startups
- What I addGo-to-market and sales acceleration
Chapter 4 · INSEAD MBA · 2026 · Wharton exchange
A break, to build with AI.
INSEAD2026 · Fontainebleau, then Wharton
#2MBA in the world, Financial Times Global MBA Ranking 2026. MIT Sloan is 1st, Wharton 3rd, Harvard 10th.
Back in school. Building again.
I left MuleSoft to step back from selling. I chose a one-year MBA at INSEAD and have followed three threads there: the classroom, a research project on AI, and building my own AI products between classes.
- 2ndof 36 teams at Master Strategist Day
- Dot.aimy INSEAD Venture Competition entry: agents that implement SaaS at scale
INSEAD · MBA 26D
Arthur Froissart
MBA 26D
Class of December 2026
Class of December 2026
In the classroom
The classes I built the year around.
- Generative AI Product and Businesswhere Reel Reads started
- Technology and Innovation Strategy
- Negotiations
- Realising Entrepreneurial Potentialentrepreneurship through acquisition
- Business and Society: Public Policy
- Startups and the Law
- Private Equity
- Mergers, Acquisitions, Alliances and Corporate Strategy
- FinTechs
- Exchange term at WhartonOctober to December 2026
A school in Nairobi, as a teaching case
Can a community school build a model that scales its impact?
- SubjectSaint Martin's School in Kibagare, which I have supported since 2022.
- WithProf. Ilze Kivleniece, INSEAD.
- StatusIn progress. Release planned for 15 November 2026.
Built during the MBA
Two products, between classes.
Neither existed before INSEAD. Each started in or alongside a course, and I built them myself in 2026.
WhatsApp
Gmail
LinkedIn
Telegram
Notes
Ambidex
Built during the INSEAD MBA · 2026
Product · built on Claude, alongside classes
Ambidex
If it can be said out loud, it can be a task.
- ProblemMy own to-do list lived in five apps and none of them talked to each other.
- BuildAn agentic voice-to-task system: speak, and it files, chases and drafts.
- StatusBuilt by me, used daily.
Built during the INSEAD MBA · 2026
Prototype · final project, Generative AI Product and Business
Reel Reads
The scroll habit, pointed at books.
- ProblemShort video is where attention goes, and almost none of it sticks.
- BuildA swipe feed where each card is one 20-second idea from a book, narrated by an AI author clip.
- StatusWorking prototype, not a live product.
That's the run so far.
If this looks like the kind of operator-builder you want close to your product, I'd love to talk.