90% of our GTM is automated. The 10% that isn't is the only reason it works.

Over the last three years I've watched a lot of companies try to remove that 10%.

Every single time, it destroyed results.

The pattern is always the same.

A founder gets excited about automation. Rips out the human layer. Replaces reps with sequences and AI agents that "do everything".

Works for a week.

Then results drop off a cliff.

Especially for high-ACV, complex offers.

So I figured I'd just write about it.

1. Automation can enrich, score, sequence, and send at scale.

It can watch a whole TAM 24/7.

It can log every call automatically.

It can build a list overnight while your team sleeps.

It cannot read a room.

It cannot hear hesitation on a call and know when to shut up.

It cannot look at a deal and feel that something is off before the data says so.

And it definitely cannot pick up the phone.

2. Cold calls are still the single highest-converting outbound channel we run.

Nothing else comes close for actually creating pipeline.

No AI agent is doing that.

No automated sequence replaces a rep who can open a conversation with a stranger in 30 seconds and book a meeting before they hang up.

That skill is human. It stays human.

3. I've seen a company burn through their entire TAM in 90 days.

Automated system blasting every signal.

No human deciding which ones were actually worth pursuing.

By the time they realised, the market was torched.

Replies dead.

Domain reputation gone.

Six months to rebuild what should have been permanent pipeline.

4. I'm not saying this as someone who's anti-AI.

I run almost everything through Claude Code.

Data orchestration. Enrichment. Signal tracking. Ad ops. Reporting.

A three-person team managing what should require 15.

But all of that AI sits behind the rep.

Not in front of the prospect.

5. The split most teams get wrong is which part is automated.

They automate the conversation.

They leave the admin manual.

That's the exact opposite of what works.

Automate the list-building, the enrichment, the scoring, the routing, the logging.

Leave the calls, the discovery, the objections, the closes.

Machines behind the rep. Humans in front of the prospect.

I share this because the "AI does everything" pitch is loud right now, and the teams buying it are the ones I'll be un-torching in 12 months.

The 10% you can't automate is the 10% that makes the other 90% worth building.

Tom