By Jordan CrawfordAI Just Took the Pants Off Your Sales Team
AI has sped everything up and it’s about to pants the reps who have been faking it.
Waddup Topliners! It’s Jordan Crawford. I write On the Edge — a top-50 Substack in under two months — and I’ve been building AI-native GTM since GPT-3.5, back when you could only use the API, and ChatGPT didn’t exist yet.
Here’s what’s true as of today (July 2026): AI hasn’t just changed what’s possible in GTM. It’s exposed what was already broken. The reps who were faking it, the campaigns that were running on hope, the “research” that was actually just procrastination. All of it. AI didn’t create these problems. It just turned the lights on.
Three things matter most right now: Where you’re running your tools, how clean your data is, and whether your team is actually shipping. Everything else is noise.
Get In the Terminal
You’re probably running Claude or ChatGPT in a browser tab. Stop.
You need to be in Claude Code or Codex. Not the desktop app, not Claude CoWork. In the terminal, full stop. Do not pass “Go.” Do not collect $200.
The difference will amaze you. Otherwise you’re living about a year in the past.
Make this one change and everything else changes.
You Can Finally Ask Your Data Anything — If the Data’s True
We can now ask questions of all our data. We’ve never had that. Which means we’re already a big step ahead.
On that basis, I’m going to make a radical statement: Reverse ICP is the way in 2026.
Instead of moving from the market down (“I sell to VPs of Marketing, super useful”), do the opposite — asking “based on everything the customer has ever said and done, and comparing that to my closed/lost and closed/won deals… what really makes a highly qualified customer?”
Two things to consider:
Your data has to be right. You’re going to spend three months on field validation, double-checking things. Go do that work. Once it’s done — or at least once Claude knows what’s wrong with it — you’ll be happy you did.
The payoff is different. It used to be a dashboard. Now, it’s better strategic decisions informed by what the customer actually said and did. Not what your reps heard, and not what dropdown someone picked in Salesforce.
Who’s Going to Get Fired and Why
Once you trust the data, the hard part shows up. You have to hold your team accountable to shipping and learning.
A client told me recently, “Jordan, it’s been five weeks and we haven’t seen the AI transformation we were promised.”
My response was “let’s look at the data.” For five weeks their team said they’d ship a campaign. Nobody shipped one. No one called.
So I said: “Report every single day on the number of calls your SDRs make.” Suddenly, they had the most productive week in company history.
There’s a lot of performative work in GTM. Teams have been hiding for a long time:
“I had to do research.”
“The lead wasn’t right.”
“The data from marketing was bad.”
Every one of those excuses raises the temperature on the next campaign. This one has to work — we’ve been on it for three months. That’s exactly backwards.
Run more campaigns, faster, and it’s okay to get a lot of things wrong. Because nothing is riding on it, as you can launch again tomorrow.
How I Describe the AI GTM Transformation
For all the bananas, here’s the entire value of AI GTM in a sentence:
AI GTM is just a way to test plausible hypotheses, with better data, anchored on real customer conversations and customer actions, at a faster clip.
That’s it. That’s the entire thing.
If you can’t ship faster, the stakes on every test go up, the value goes down, and so does your learning. But if you stop being precious about any single test — if you ship more of them with faster feedback — you can actually steer the business. No single thing has to carry all the weight.
Stop Targeting by Your Tools
AI unlocks ways of combining and structuring information. In this new paradigm, it’s your imagination that’s the bottleneck, not the tooling.
Here are examples to expand your imagination. I built all of these in weeks:
1. Find and score new innovative channels
Ask the question: Where are your buyers required to show up?
What it does: Score every channel by CPC, reach, and fit. (There are hundreds you may not even be considering.)
What I did: Evaluated ~50 channels for a company targeting lawyers to find the best CPC. The winner was a bar association with compulsory membership = 100% of a market.
2. Sort your market by influence
Ask the questions: Who shapes your market? Who are the influencers in your category by share of voice with your customers (e.g., podcasts, speakers, etc.)?
What it does: Determine which customers actually carried weight in their industry.
What I did: Sorted 1,145 customers to find the 56 most-connected; the same engine ranked ~15,000 U.S. fire chiefs to a top 200.
3. Filter doctors by what they prescribe
Ask the questions: Among medical specialists in a given territory, who prescribed what, how did they bill, and then, where do they work?
What it does: Find the medical professionals who can prescribe important medication for patients before the healthcare crisis.
What I did: Sorted 9,668 cardiologists using Medicare data on who treats amyloidosis or HFpEF.
4. Score all sales reps you could hire by who they sell to and their prowess
Ask the question: Among reps in your market, what are their deal sizes, how fast did they get promoted, do they mention awards, etc.?
What it does: Screen top-tier reps against context, then success.
What I did: Scored 15,489 reps down to the 135 who sell like you, for about $10.
5. Find emails on 100K+ websites for free
Ask the question: How can I find contact information for a key demographic without having to pay an arm and a leg?
What it does: For those going after SMBs, this downloads an entire market’s worth of websites and pulls emails from anywhere on the page.
What I did: Pulled named contacts from two whole markets for $0.
6. Find every parking lot in the country, who owns it, and score it by pothole count
Ask the question: Which facility is in most need of our services (based on specific problems that we can solve)?
What it does: Finds every parking lot from an open source version of Google Maps, uses free flyover imagery data, and takes a cheap LLM to score them all by the investment it would take to fix them for a national paving contractor.
What I did: Scored every parking lot in America for $45, ranked by the amount of damage.
7. Find every franchise and franchisee in the country
Ask the question: Who are the franchise owners who operate more than one location, nation-wide?
What it does: Every franchise in the country is required to register, and their data is stuck in huge PDFs.
What I did: Mined 142,579 franchise operators out of legal filings that nobody ever reads.
8. Found the most advanced Claude users in the organization
Ask the question: Who’s actually shipping with Claude vs. just chatting with it?
What it does: Exports all the prompts from your Claude Enterprise plan and scores who’s asking the best questions.
9. Replace an $18K/mo agency with two weeks of Claude Code training
Ask the question: If you weaponized your own domain expertise in Claude Code, what could you build that an agency can’t?
What it does: Teaches a non-technical SME to ship SEO work directly – no middleman required.
What I did: Taught one SEO expert Claude Code; two weeks later she was up +20% organic, +22% CTR, #1 AI-cited tool in her category, and fired her $18K/month agency.
Your Real Job: Find the Fulcrum
Notice what every one of these examples has in common. Each took a day and pocket change to accomplish. Neither the data, nor the build, nor the budget were the bottleneck. And every one of them ends the same way — dead on the page until a human picks it up and acts.
When you can build anything in an afternoon, the only variable left is whether your team ships. That’s the dirty little secret: in many cases, this stuff is going to generate more work for you and your team.
Your job now is to figure out which seesaws to bet on, and how to move the fulcrum.
AI will help you differentiate between the things you should stop doing (i.e., the stuff not moving the needle), the messy middle (i.e., the things that kind of make sense but give you no real leverage), and the few things that actually move the business.
Once you know that, the only question left is: can you enable your team to test 20 of those ideas at rapid iteration, with fast feedback from the market?
And the feedback loops are faster than you think. GEO and SEO: launch a change, Google indexes it, and you can watch ChatGPT start citing you in days or weeks. Cold calling: feedback in days — change the list, call again. You can even hash your prospects’ information and upload it as an ad audience, so you’re running ads only to the 200 most influential fire chiefs in the country.
Release Your Imagination
To do this as a leader, you have to unlock your imagination from your old tools, ask bigger questions, and learn faster. That means shipping to production, lowering your quality bar, and testing in low-stakes environments.
If a rough test fails and you learn in a day — and the stakes were “a small experiment,” not “delete the customer database” — who cares? You know if it worked.
It also means holding your team to a standard of shipping that probably means layoffs, because people have been hiding for a long time. And all the hand-wringing about hallucinations and quality melts away when you ship and test fast in places where a miss is cheap.
Where This Leaves You
You’re a revenue leader. You’re in the best possible seat to have the biggest imagination in the building. Set the fulcrum where one win pays for everything. And make sure your team can run the tests at lightning speed.
Everything else, right now, is noise.
Thanks for reading. Your attention is a precious thing in 2026, and I’m grateful for it.
Jordan Crawford is the founder of Blueprint and writes On the Edge, a top-50 business Substack. He’s been building AI-native go-to-market since GPT-3.5, has advised Clay since 2020, and has crossed 10 billion tokens on OpenAI. He teaches revenue teams to build their own data and ship at the speed of imagination.
