Cases
What I have built and the result it had, told the same way each time. If a fact has no evidence, it is not here.
Cases and results
These are cases from my roles at Sesame HR, Dentsu and PHD, and from my own projects. Each one says what kind it is. Freelance client cases will be added once each client agrees to publish them. Figures are relative: absolute ones belong to each company.
Real implementation
AI GTM Engineering
Sesame HR, in my role as Head of AI GTM Engineering, since April 2026. It is an internal project, not a client's.
- Context and problem
Marketing, sales and data work is spread across five markets.
- How it was done before
It depended on manual steps and on the few people who knew how to do them.
- What was built
I designed and run a system of AI agents with ten-plus tools in production.
- How it works
The agents qualify leads, watch campaigns, prepare content and answer data questions.
- Result
Ten-plus tools in production across six areas and five markets.
- Evidence
How the system is organised is on “About”, under “What a programmable organization looks like”.
- Limits and the human role
People approve what matters.
Real implementation
Paid media and marketing operations
Sesame HR, in paid media and marketing operations, 2023 to 2025. It is an internal project, not a client's.
- Context and problem
Sales opportunities had to grow in five markets at once.
- How it was done before
Reporting, tracking and lead handling did not keep up.
- What was built
Lead scoring and account-based programmes. Executive dashboards that build themselves. Privacy-first tracking.
- Result
Sales opportunities grew month on month and quarter on quarter. Within thirteen months I was leading paid media and marketing operations, and I was named the company's MVP of 2024.
- Systems
The ad platforms and the CRM.
- Evidence
The MVP award, chosen among 370 employees. Growth figures belong to the company and are not published.
Real implementation
Search automation for Apple and Square
Dentsu International, London, 2019 to 2020. First I ran search accounts, then I was Global Lead, Automation & Tech.
- Context and problem
I ran search for Apple worldwide and for Square in the UK.
- How it was done before
Search accounts were checked and reported by hand, account by account, across the network's teams.
- What was built
Automations, dashboards and best practice rolled out across the network. I also trained the teams.
- Result
+47% conversions year on year on the accounts I ran, with cost per acquisition still improving. And +35% return on investment on Le Creuset's global account.
- Evidence
Account results for 2019 and 2020, and my manager's recommendation on LinkedIn.
Real implementation
Search planning for Unilever
PHD Media (Omnicom), London, 2017 to 2018.
- Context and problem
International search campaigns for Unilever and Upfield.
- How it was done before
Reports for the EMEA teams were put together by hand.
- What was built
I automated the reporting. I also planned and optimised search for both brands.
- Result
+221% conversion rate year on year.
- Evidence
Account results for 2017 and 2018. The client's recommendation is on my LinkedIn profile.
Own project
GeoAdsFinder
Own project. It started during my time at Dentsu International and is a standalone product today.
- Context and problem
SEO and paid search teams need to see what someone searching Google from another city or country sees. At Dentsu it was a problem the whole network had.
- What was built
A tool that shows what someone searching Google from any city or country sees, ads included.
- How it works
It also lets you see competitors' ads on Meta, Google, TikTok and LinkedIn. The basic version is free.
- Result
The Dentsu network adopted it. Today it is a standalone product, adding visibility in AI search.
- Evidence
Still live at geoadsfinder.com.
Own project
#YoMeTesteoEnCasa
Own project with two friends, in March and April 2020.
- Context and problem
When COVID-19 arrived, there was no self-assessment tool for the whole of Spain.
- What was built
The first COVID-19 self-assessment tool for the whole of Spain, built in 24 hours.
- How it works
A chatbot guided each person and gave them the emergency contacts for their area.
- Result
Local authorities and hospitals backed it.
- Evidence
Expansión and Xataka, among others, covered it.
Keep reading
- How I work
The process step by step, what you pay for and which work fits.
- Services
Implementation, diagnosis and training: who it is for, what you get and how it starts.
- About
My background and how I think about programmable organizations.
Do you have a similar task?
Tell me which task takes up most of your time and how it is done today. I'll tell you where I would start.
Tell me which task to take off your plate