CATCH PROBLEMS WHILE THEY’RE STILL SMALL.
The project looked fine.
Status: green.
The weekly update said everything was on track.
Then one week, it wasn’t.
A task had been slipping by a day here and there.
One person had stopped replying as quickly.
A dependency was sitting with another team.
Nobody thought any one of those things was a big deal.
Then they all showed up at once.
Looking back, the warning signs were there.
They just never became a warning.
LET’S FIX THAT IN 12 MINUTES.
STEP 1 — GATHER THE SIGNALS
That’s too broad.
Give AI the evidence you already have.
Your recent updates.
Task lists.
Meeting notes.
Open items.
Anything that shows what’s actually happening.
🔴🟡🟢 COPY THIS INTO CHATGPT OR YOUR AI TOOL:
Act as my project chief of staff.
Review the project information below and look for early signs that something may be drifting or becoming a problem.
Look for:
- Tasks that keep slipping
- Items with no clear owner
- Work waiting on someone else
- Repeated blockers or unresolved questions
- Dependencies that haven't moved
- Changes in timing, scope, or priorities
- Anything that was expected to happen but hasn't
For each signal, show:
- What you noticed
- What evidence supports it
- What could become a problem if nothing changes
Rules:
- Do not invent risks.
- Separate what the information shows from what it might mean.
- Don't treat one small delay as a major problem without evidence.
- Keep it concise.
Project information:
[paste your updates, notes, task list, or other information here]
Now you have a list of signals worth looking at.
Not a prediction.
Not a disaster report.
A short list of things that deserve attention.
STEP 2 — SORT BY WHAT COULD HURT
Now take each signal and ask:
What should we do about this?
You don’t need to escalate everything.
You need to separate normal project noise from something that needs attention.
STEP 3 — WRITE THE WARNING
For every ACT NOW item, turn the signal into one plain sentence.
That’s much more useful than:
“I’m worried about the timeline.”
Now someone can actually do something.
WATCH FOR THIS PHRASE:
“It should be fine.”
Maybe it will be.
But if something has slipped three times, has no owner, or is waiting on a dependency nobody is watching, “should be fine” isn’t a status.
It’s a guess.
Name what you’re seeing.
Then decide whether it needs action.
SEND IT
Share the short list with the people who can actually act on it.
Keep it simple:
ACT NOW
What’s at risk
Since when
What needs to happen
Who needs to act
WATCH
What you’re watching
What would make it move into ACT NOW
Save it:
Early Warning Check v1
Now the warning isn’t sitting in your head or buried in a meeting.
Someone can see it.
Someone can act on it.
HOW DID IT GO?
WHAT YOU JUST DID
You didn’t wait for the project to become a problem.
You looked for the small signs that something was starting to drift.
A late task.
A missing response.
A dependency that isn’t moving.
A decision nobody has made.
None of those automatically means the project is in trouble.
But they give you something to check before the problem gets expensive.
Before you grab today’s Field Kit:
Tell us about your role and what you’re working on so we can build workflows around the problems you’re actually dealing with.
The warning was heard.
The right people saw it.
And then…
Nobody did anything.
That’s the next problem.
A warning that says “something feels off” is easy to hear and easy to forget.
Thursday, we’ll look at how to turn a warning into something specific enough that someone can actually own it.
TODAY’S FIELD KIT(EARLY WARNING CHECK):
Inside The Kit:
The full Early Warning Check prompt sequence
Signal tracker
One-line warning template
This kit lives in our learning community. Sign up free to get access.
