Most executives I speak to right now are stuck in what I call the PoC Lobby.

Months are spent evaluating models, running pilots, and building “AI roadmaps”…
Only to end up exactly where they started, with no measurable impact on delivery, cost, or speed.

The assumption is that the technology isn’t ready.

It is.

The problem is execution.

---

The uncomfortable reality

Based on recent work across infrastructure and project-led organisations, the pattern is consistent:

• Teams are still rebuilding reports manually every week
• Planning cycles are slowed by fragmented or unreliable data
• AI initiatives exist, but sit outside actual delivery workflows

In other words:

AI isn’t failing.
It’s just not being applied where the real friction is.

---

Three things that actually matter

From 50+ real-world deployments, three truths show up every time:

1. Most of the work is invisible​
The challenge isn’t the model, it’s data quality, process design, and adoption inside live teams.

2. Iteration beats planning​
Every successful implementation moves quickly from idea → test → adjustment.
Not 6-month strategy cycles.

3. Productivity gets worse before it gets better​
There is always a dip while processes are restructured, and most organisations abandon efforts at exactly this point.

---

Where this breaks down

The biggest gap I see is this:

Organisations treat AI as a strategy exercise​
When it’s actually an execution problem

So you get:

• 12-month roadmaps
• Slide decks
• Pilot use cases

…but no change to how work actually gets done.

---

What works instead

The teams that move fastest do one thing differently:

They take a single operational bottleneck​
…and build something usable around it immediately

Not perfect. Not scaled. Just working.

In practice, that looks like:

• Turning a manual reporting process into a live model
• Stress-testing schedules using existing data
• Matching tender requirements to team structures automatically

Not theory — usable outputs.

---

Where this becomes real

For most teams, this shows up today as:

• Reporting cycles rebuilt manually across multiple systems
• Delays in decision-making due to lack of visibility
• “AI initiatives” that never move beyond pilot stage

This isn’t a strategy issue.
It’s an execution bottleneck.

What we typically do is take one of these friction points and turn it into a working model in a day — using the data and tools already in place.

If that’s something you’re currently dealing with, just reply with:

REPORTING​
or
​PLANNING

I’ll share how we’re approaching it in practice with other teams.

---

Muriel Demarcus
CEO, Marsham Edge