Automation should always be prioritized against delegation.
It forces precision. Hence, it increases efficiency.
What can't or shouldn't be automated will appear even more clearly.
Your AI thinks your biggest customer is a cold lead.
It’s about to cost you the renewal.
Picture it. Your top closer quits in Q4.
Their context lived in their head.
Their CRM notes were vague at best.
The new rep inherits an empty record.
Your AI agent reads that record, sees nothing, and fires off a cold intro to a customer who signed a six-figure deal last year.
That’s not an AI problem. That’s a data problem.
I see it very often in real life.
The most valuable context in a company almost never lives in the system. It lives in one person’s head, one Slack thread, one notebook nobody else opens.
The system has to hold the context. Not the person.
This is the real case for GTM automation. Not just efficiency.
Continuity.
Every signal, every touchpoint, every note lives in the system. So the next departure doesn’t empty your pipeline.
Your reps will leave. Your context shouldn’t walk out with them.
Name the one piece of deal context in your org that only lives in someone’s head right now.
That’s where the leak starts.
Brain drop after @Clay’s conference in London today (in no specific order):
- This is only the beginning of the story
- It started with data enrichment, it continues with orchestration
- Some companies are doing GTM engineering and having amazing results, while others (most of them) don’t know what an API is
- Clay’s marketing and branding are the best of the game
The real cost of Claude Code isn’t tokens or API credits.
It’s the deals you didn’t close.
I get it.
Claude Code is exciting. I’m first in line to build with it.
But I’m watching people go crazy right now.
Ripping out Clay. Ripping out every SaaS tool in their stack.
Rebuilding everything from scratch.
It feels like progress. It isn’t.
By the time you’ve rebuilt the tool you already had, fought the errors, and signed up for the maintenance forever, your pipeline hasn’t moved an inch.
That’s the trap.
If your pipeline isn’t growing while you build, you’re not making progress. You’re building technical debt.
The GTM engineers winning right now aren’t the ones rebuilding the most. They’re the ones who know what NOT to build.
Buy the boring infrastructure.
Build only the edge that actually wins deals.
If you want to launch a new tool, that's a full time job.
If not, focus on what matters the most for your business.
If you still assume the CRM is a place where reps should manually report, you're in troubles.
Nobody likes filling out another field. Another dropdown. Another note.
Don't be surprised when the data is incomplete.
If the system depends on humans doing low-value admin perfectly, the system is fragile by design.
CRM inputs should be fully automated:
→ Calls get summarized.
→ Signals get captured.
→ Accounts get enriched and monitored.
→ Fields get updated.
→ Next steps get created.
→ Risks get flagged.
These are (crucial and needed) low-skilled tasks.
That's AI agents work.
Biggest update to my Ultimate Claude Code guide for GTM since launch.
Major update. Still free.
What's new:
→ More connectors. And the right ones.
The tools to plug Claude Code into your revenue stack, ready to wire.
→ Deployment, done properly.
There are several ways to ship a Claude Code agent in production. Which one is right depends on your team, your stack, and your context. The new version walks through all of them, with a decision tree.
→ New tools to make your agents sharper.
Each one a small upgrade. Together they compound.
The original 4 sections, 42 topics, and 47 ready-to-use skills are still in.
Not a prompt library.
A complete OS for building GTM automation with Claude Code.
If you want it:
1. Like this post
2. Follow me
3. Comment "Claude" and I will DM it to you
4. ♻️ Repost so other GTM teams can find it
AI did not kill expertise.
It killed average.
Before, being decent at writing, research, analysis, or execution was valuable because the baseline was low.
Now the baseline is much higher.
A junior person with AI can produce work that looks "pretty good."
A mediocre operator with AI can move faster.
A below-average professional with AI can hide for a bit longer.
But only for a bit.
Because when everyone has access to the same average machine, average stops being impressive.
The value moves up.
To taste.
To strategy.
To judgment.
To knowing what matters.
To understanding context.
To knowing what to ignore.
To making the right tradeoff.
That's where humans become more valuable.
If you are below the AI average, you are in trouble.
If you are just above it, that becomes the new normal.
If you are in the top 1% of expertise, AI becomes leverage.
Alex Hormozi recently dropped a video about AI agents.
He nailed the principle:
Stop thinking in roles, start thinking in workflows.
He didn't give the playbook.
My next video is the playbook.
Most companies are still trying to replace roles with AI agents. "AI SDR." "AI Marketer." Wrong frame.
Your org chart was built to coordinate humans. Agents don't need a manager. They need a task.
Take any role. List every task that person actually does in a week. Every audit I've run lands 60 to 70% of them not needing a human at all.
That's how a small team outproduces a fifty-person org.
We're building TC9 this way. Output of ten people, fraction of the headcount.
The full breakdown drops today on YouTube.
The 60% is the floor, not the ceiling, and there's one rule that decides which tasks stay human.
Watch it here: https://t.co/ttsnkECQHS
There is a perfect negative correlation between outbound volume and reply rate.
lemlist pulled the data. The math gets interesting.
Here's what Charles Tenot, CEO at lemlist, shared with me when I interviewed him:
They split their customer base into 10 buckets by reply rate. 2,500 customers per bucket.
→ Top decile = best reply rates = sends the LEAST volume
→ Bottom decile = worst reply rates = sends the MOST volume
→ Perfectly inverse. No exceptions.
The logic is obvious in hindsight.
Less volume per rep means more time per message. More time means more relevance. More relevance means higher reply rates.
So "personalize more" isn't the advice. The real advice is "send less."
But here's where most operators stop reading. And it's where the actual insight starts.
Charles dropped this:
"If you do 10x the volume but divide the reply rate by 5, you still have 2x the replies."
That single line flipped my entire model of outbound strategy.
We spent half an hour together to discuss Sales, Growth, and AI. Check it out!
Full interview on YouTube: https://t.co/x4MkbhpqxE
3 tiers in cold outbound right now.
Most teams are still playing the wrong one.
→ Bottom: Noise
Mass templates with no relevance. Just skip this.
It's just bad work that (too many) companies still pay for.
→ Middle: Relevance
Right timing, right signal, right account.
This used to be a craft. Now it's the new norm.
Anyone serious has either automated this layer or is about to.
→ Top: Alpha
The unique insight on the account or the person that only you found.
A pattern no one noticed.
An anticipated signal before the obvious one.
A read on why their stack choice creates a problem they haven't named yet.
This stays human.
So outbound just split into two real jobs:
Relevance. And Alpha.
→ Use AI and automation for relevance.
→ Use humans for alpha.
The teams that win in 2026 will build agents for the relevance layer, then move their best people to mining alpha.
Everything else is wasted pipeline budget.
Drop the last piece of alpha you put in a cold email that an AI couldn't have written!
Cold email is like street advertising.
Most senders haven't figured this out.
When I interviewed Charles, CEO at lemlist, he put it perfectly:
"You see thousands of ads in your life. But one sticks, and you think, 'Wow, this is interesting.'
Outbound is the same. You receive hundreds of emails. 90% is out, you don't like it. Maybe 1% will be, 'Oh, that's interesting actually.'"
That's the bar.
→ Not "every prospect should reply."
→ Not "personalize harder."
→ Not "find the magic subject line."
The job is to build the 1% that sticks. Everything else is noise, and that's fine.
The best billboard on the highway isn't the one with the most words. It's the one you remember 3 hours later.
Your cold email should pass the same test.
We went deeper on this for half an hour: where the 1% comes from, what most teams get wrong when they scale volume, and how AI shifts the math.
Full interview on YouTube: https://t.co/x4MkbhpqxE
Your highest-intent buyers visited your site yesterday. You’ll never talk to them.
97% of your website traffic leaves without filling a form.
The 3% that does gets nurtured for 6 months. By the time that sequence ends, the rest have already bought from someone else.
That math is what broke the funnel.
The modern playbook flips the order:
→ Generate traffic (same as before: SEO, ads, social)
→ Identify the individuals behind the traffic
→ Match them to your ICP in real time
→ Hot ICP buyer: fire an intent-led outbound sequence (email + LinkedIn DM) the same day
→ ICP but not ready: enrich, score, and watch for hiring or funding signals
→ Not ICP: push into the classic content funnel
The old funnel still exists.
But it’s now the fallback path, not the main one.
Three things changed to make this work:
1. Visitor identification tools resolve people
2. Signal workflows qualify in real-time
3. Outbound platforms fire sequences in minutes
Your highest-intent buyers were on your site yesterday. They didn’t fill a form.
Time to get your marketing plays to the next level.
He scraped 75,000 leads in 20 minutes.
No engineer. No agency. Just Claude + Python.
When I asked Charles Tenot, CEO at lemlist, about AI in sales, this is what he told me:
"Last week I built a Python script with Claude, and I scraped a database of leads in 20 minutes. 75,000 leads.
Before, that required a growth engineer, and you were struggling to find one that was good.
Now any company can do things they were not able to do before. You can scrape databases, visit websites, do activities that humans would not have done because the ROI was not good enough.
All of that can be solved with AI. You can have 10X data quality. And that's at the fingertip of everyone."
And that was the warm-up.
We spent an hour talking about what works in Sales today (supported by real life data), how does the future look like with AI, the role of humans, and so much more.
Full interview on YouTube: https://t.co/x4MkbhpqxE