Hiring AI advice? You're hiring a translator.

Most of the room talks about AI in a language you cannot spend money on. I answer in yours, and prove it with working systems, not decks.

Get the weekly insights

Free. One email a week. Unsubscribe anytime.

Read the research first

I have been making technology legible for twenty years: first for retirees who had never touched a computer, then as founder of a software house shipping secure systems for Orange and EDF.
Today, an independent advisor to executives carrying a question no one else in the room wants to admit they cannot answer.
What do we do with AI first, in what order, and why?

What I do

Show, don't tell they say.

I build a working system on your own data. You leave knowing what is possible.

Sequence

A paid audit, I interview your team and review your key processes. I deliver a plan as a proper business proposal.

Carry

Ongoing counsel, strategy, prototypes, working systems, until your team runs it without me. That's the exit criterion.

Building momentum and handing over the wheel back to you

Why I work this way

I run experiments before I give advice. Vendor demos are built to impress. I run systems long enough to find out what breaks.

The only way to know whether AI works is to run it. So this practice publishes its failures.

Reports, not slides: research-grade documentation you can actually use.
No vendor alignment: I work for you, not for an AI platform.
Confidentiality first: your data stays yours, and nothing gets published without your sign-off.

The experiments, postmortems and failures are public

The long version

The translator's thread

1994

Found a BASIC programming book for the MO5. Where it starts.

1996

My first computer. Disassembled it 3 months later to understand how it worked.

1999

Learnt C and C++. Small 3D games with OpenGL from 2002.

2003

Building and selling PCs in a computer shop in Grenoble during week-ends.

2004

Teaching computer basics to retirees. My first translation job, making something opaque usable for people who felt locked out.

2005

Built Linux from scratch (LFS Project). The discipline of knowing every layer you stand on.

2007

Graduated from ENSERG (Grenoble INP). Then years on embedded systems and embedded Linux at Freescale.

2014, for a decade

Co-founded a software development agency and led its 15 people. On-premises, secure software for broadcast (Orange / Globecast) and energy (EDF). An industry where "it works" is not an acceptable answer.

Today

Advanced Stack. I help companies figure out whether AI can actually work for them: working prototypes on their own data first, advice after the evidence.

Who I work with

For

Ambitious companies looking to make a difference, strongly biased towards humanism
Decision makers who needs the AI agenda defined and carried - without hiring an in-house team yet.

Not for

Companies whose AI goal is cutting headcount, we don't take those projects.
Pure execution work, there are excellent IT services firms for that.

The person behind Advanced Stack

Pierre Alexandre Schembri

··

I am the one in the demo, the audit and the field notes. No juniors handed to you after the pitch, senior from day one.

When capacity is the constraint, I partner deliberately with senior consultants. It stays my practice and my judgment.

You've read this far. The next step costs one email.

Subscribe to the weekly insights