Marco Domeniconi ← The Carnet
Fine-line engraving of a centered Roman sundial on a columned plinth. A finely drawn summer sun sends a thin ray toward the gnomon. Faint hour lines fan across the dial, and the cast shadow is drawn in red, the single true reading. Folio 04 of the Carnet.
CARNET· Folio 04· A6 · A5 · Perform & Transform

The quiet of summer is not a gap in the calendar.

It is the one instrument you are handed all year.

The goldrush has a single grammar: do the same work, faster, by handing every desk a tool. It is pure Perform. More speed, same system.

The numbers say speed is not becoming progress.

MIT NANDA found that 95% of enterprise GenAI pilots produce no measurable return, on 30 to 40 billion dollars spent. High adoption, low transformation. They named it the GenAI Divide.

Gartner expects over 40% of agentic AI projects to be canceled by the end of 2027, for costs that overrun and value that stays vague.

In Switzerland, only one in three AI pilots reaches scale, while Swiss users report drawing more from AI than the global average. So the brake is not the tool, and not the talent. It is the passage to scale.

That passage is a system problem. A system has a throughput, fixed by its constraint, not by the velocity of its fastest desks.

You do not lift a constraint by adding tools. You lift it by redesigning the system around it.

Summer is when the FOMO noise falls quiet and the constraint finally becomes visible. Distance is the condition of that sight, not its interruption.

So four steps back, before the autumn wave.

  1. Distance. Read the throughput of your people, not the speed of the fastest.
  2. Constraint. Name the constraint before accelerating the quick, and let your data make the bottleneck legible.
  3. Redesign. Redesign one role around an intelligent agent that has proven itself, instead of adding one more tool to every desk.
  4. Re-entry. Install a re-entry gate as governance: this autumn, no pilot ships without proof of system throughput, held by a named owner.

Do this, and the autumn wave will not reproduce the 95% and the 40%. Skip it, and it will.

Autumn will not reward those who deployed the most. It will reward those who stepped back the best.

This communication was researched, written and illustrated by AI agents (Livia, Aristide, Romeo) with separated roles, under human supervision, with final sign-off by Marco Domeniconi. Producing it took about two hours of orchestrated agent work and roughly 3 million tokens, for an approximate cost of 15 CHF.

Sources

  1. MIT Media Lab, Project NANDA, The GenAI Divide: State of AI in Business 2025 (July 2025). mlq.ai
  2. Gartner, Over 40% of Agentic AI Projects Will Be Canceled by End of 2027, 25 June 2025. gartner.com
  3. ICTjournal, En Suisse, deux tiers des projets IA pilotes restent au stade expérimental, 9 June 2026. ictjournal.ch
  4. ICTjournal, Les salariés suisses tirent davantage profit de l'IA que la moyenne mondiale, 13 July 2026. ictjournal.ch
← Back to the Carnet Marco Domeniconi on LinkedIn