Field Notes

If 2025 was the year teams “played” with AI, 2026 is the year it exposes who actually understands systems.

Because the uncomfortable truth is this:

You can add AI to a broken growth engine…
…but you’re just shipping the same dysfunction faster.

I’ve seen this pattern repeatedly: a company proudly says, “We’re using AI everywhere,” while the basics are still missing:

• no shared definitions

• no consistent follow-up

• no visibility into where leads die

• no feedback loop between departments

• no governance for what should (and should not) run automatically

So before we ever talk about automation, we run a simple operating reset.

Not a “digital transformation.”
Not a tech overhaul.

A 90-day reset (time frame depends on the infrastructure complexity) that turns guesswork into an operating model.

The premise

Most “marketing problems” are systems problems.

If you can’t answer what happens to £1 when it enters your funnel, you don’t have a growth engine.

You have activity.

The 90-Day Reset (how it actually works)

Days 1–15: Map the real system (not the slide deck)

You follow one lead end-to-end:

first touch → qualification → booking → close → onboarding → retention

Then you document:

• every handoff

• every decision point

• every place a human is acting as a slow API (copy/paste, chasing, guessing)

In almost every business, this reveals the same thing:

The “funnel” is actually 12 disconnected workflows and one heroic operator holding it together.

Output of Days 1–15:
A single page that shows where the system truly leaks.

Days 16–45: Stabilise the infrastructure (plumbing beats cleverness)

This phase is boring by design.

You standardise:

• lifecycle definitions (what counts as a lead, warm lead, qualified, opportunity)

• pipeline stages and entry/exit criteria

• “single source of truth” rules (where reality is recorded)

You remove or reduce:

• duplicate tools doing the same job

• shadow spreadsheets and backchannel “truth”

• ambiguous ownership (“someone should follow up”)

Rule:
If the system is chaos, AI will produce chaos—faster.

Output of Days 16–45:
A stable operating model that the team can actually run.

Days 46–75: Layer intelligence (narrow, high-leverage)

Only now do we introduce AI—carefully.

One use-case per department is enough:

• Marketing: research + drafting support

• Sales: summarisation + next-step suggestions

• Support: triage + suggested replies

This is not about “wow factor.”
This is about adoption and accuracy.

The tell:
If your team quietly stops using it, it’s not leverage.

Output of Days 46–75:
AI becomes a disciplined assistant, not a random experiment.

Days 76–90: Automate the obvious (autopilot is earned, not promised)

Automation comes last.

We automate:

• reliable patterns

• repeatable work

• low-risk tasks with clear boundaries

We do not automate:

• sensitive outreach

• pricing decisions

• high-stakes approvals

• anything that could damage trust without supervision

We add governance:

• approval gates

• thresholds

• rollback rules

• exception handling

Output of Days 76–90:

A supervised system that can scale without breaking.

Why this works (and why most teams don’t do it)

Most teams want AI to be the solution.

But AI is not the solution.
The operating model is the solution.

AI is just what happens after the model is stable.

And stability requires something founders often avoid:
definitions, discipline, and ownership.

The one question that tells you if you need this reset

If you can’t answer:

“What happens to a lead the moment it enters our business—every time?”

Start here.

Closing note

The companies that win in 2026 won’t be the ones with the most tools.

They’ll be the ones with the strongest operating system.

"Skills create income. Systems create scale. Operating Systems create freedom." Maris Spalins.