Your demand and outbound read like everyone else's AI
Buyers screen generic on sight now. The fix is not less AI, it is a system that carries your voice and your account knowledge into every first touch, in each market you sell into.
The 90-second version
The generic-AI version of your demand and outbound is quietly losing the reply, because every team now runs the same good-enough AI on the same job and the output converges.
Buyers screen generic on sight, and this week a public detector put a number on how obvious the AI tells are.
So capability stopped being the edge. Your voice, your account knowledge, and your judgment are what a shared model and a detector cannot reproduce.
The European read: your buyers in every market can smell a templated first touch, and in a language that is not their first, generic reads even flatter. Sounding specifically like you, in their market, is what still earns the reply.
The thread: capability is now a commodity. Your voice, your account knowledge, and your judgment are the edge.
Read time: ~7 min.
This week’s number 1 move
Before you send another AI-assisted email or post, run it through 2 tests: would your buyer know it came from you, and could a competitor have sent the exact same words.
Demand and outbound are where you turned AI volume up first, so they are the first place buyers learned to skim. When every team runs a good-enough model on the same job, the output converges, and the generic version quietly loses the reply you used to get. That cost shows up in your reply rate, not your tool bill. Two things this week made it concrete. Buyers now pattern-match AI on sight, and a public detector put a number on how obvious the tells are. Neither is the story. The story is that your first touch has to sound like a specific operator in a specific market, or it gets filtered before anyone reads your offer.
My take: I spent 3 years and thousands of prompts explaining who I am to a model, and it still read like anyone. What changed was not a smarter model. It was feeding it my own work and keeping my judgment on top. That is why this shift lands the way I already learned it the hard way: the model stopped being the moat. When every team runs the same capable AI, the lean European team with a real voice and real account knowledge out-converts the big team blasting generic volume. The edge is not what you can generate. It is what only you would say to that buyer, in that market.
Do first: Take your last AI-assisted post or outbound email and run 2 tests: would a detector call it AI, and could any competitor have sent the same words. Any yes means your voice is not in the system yet.
The rule: If a detector would flag it or a competitor could have sent it, it is not done.
Sources: Ruben Hassid, “Can you detect AI?” (July 22) · Ruben Dominguez, “Anthropic just cut the price of frontier intelligence in half” (July 24)
Hi, it is Koen Stam and welcome to GTMcraft OS: The Future GTM Operator. This newsletter is built from 100,000+ GTM signals collected from 100+ operators and founders, combined with 13+ years of my own lessons and failures from the trenches. I write at the intersection of go-to-market practice and AI-powered systems for founders and GTM operators scaling their next 2M, 5M or 10M ARR.
100,000+ GTM relevant signals from LinkedIn, Newsletters and Podcasts indexed. Translated into 100+ GTM plays, skills and training for you to implement today.
13+ years over 3 GTM operator jobs across 3 GTM motions (SMB, MM, ENT). Scaling from 2-10M+ ARR multiple times. Same recipe. Different motions.
Now all part of the GTMcraft Operator System.
Problem 1: Your content engine scaled, and now your demand reads like everyone else’s AI
You wired AI into your content to ship more demand, and it worked, until every post or email started sounding like every other AI post/email. Ruben Hassid put the stakes in front of us: Substack now runs a detector to flag AI writing, and the tells it screens on are the same ones your drafts carry, stock openers, the tidy 3-part summary, words no operator says out loud. Charlie Hills shows the fix at full scale, and it is the most stealable play of the week. His content system does not just prompt for volume. It drafts in his voice, scores each draft against a rubric built from his own best and worst past posts, and only then hands it to a person. He runs 80% of his output through it and hired 3 people to stage and publish while he stays the final gate. The lesson for your demand engine is not to slow down. It is to make your voice a component the system reads every time, plus a human gate, so scale and sounding like you stop being a trade-off.
My take: For 3 years I explained who I am to ChatGPT (before I switch only this January to Claude, and maybe are moving back…) every time and still got the internet average back. The week it broke open, I dropped in 20 of my own posts and 2 playbooks, then pasted in a raw voice note from a walk, and what came back finally read like me (and I repeat this at least every quarter). Demand is where this bites first, because content is where most lean teams turned the volume up. I would rather ship 3 posts that could only have come from us than 10 that could have come from anyone (this sounds easer then it is to be honest). Feed it your best work, keep the last 20%, and the voice stays yours (but amplified by AI).
The European read: In a home market your audience knows your voice, so generic content reads as a step down. In a new market you are building that voice from zero, and generic AI content is how you stay invisible. A voice-loaded system is how you sound like a local operator in each market instead of a translation of everyone else.
Do first: Pull your 3 best-performing posts and write the 5 lines that make them sound like you and not generic AI. That is the start of your voice file.
Do this week: Have your AI draft your next piece against that voice file, then add 1 gate before it ships: does this sound like us, and would a detector call it AI. Put 1 human approval step in front of anything that publishes.
Do this month: Score new content ideas against your best-performing past posts before you write, so effort goes to what your audience already rewards.
The rule: If a detector would flag it or a competitor could have posted it, run it back through your voice file before it ships.
You know it worked when: Your team ships demand content in your voice without you writing it, and engagement holds or climbs.
The play: 5 steps to make Claude write in your voice (2026 series). Loads your voice into a content system so scale and sounding like you stop being a trade-off. Best for 0-5M founder-led demand engines. Fully inlined below.
Sources: Charlie Hills, “I made myself replaceable (and hired 3 people)” (July 19) · Ruben Hassid, “Can you detect AI?” (July 22)
Problem 2: Your AI outbound reads like AI, and buyers filter it before they reply
Every rep bought the same AI and pointed it at the same job, so inboxes fill with first touches that all open the same way and say nothing specific. Ruben Hassid’s detection piece is the warning shot: the tells that flag AI to a machine are the ones a buyer clocks in half a second, and reply rates fall as prospects learn to skim past the pattern. The generic AI email is not personalization, it is volume wearing a costume. The fix transfers straight from Charlie Hills’ content system: score every draft against what actually performed, so nothing generic gets through. Strip the AI tells, then load one real detail per account, the trigger, the number, the line their own team would recognize, so the first sentence could only have gone to that buyer. A cheap model can research that fast now. What it cannot do on its own is sound like your best rep on a good day.
My take: Buyers do not run a detector, they have become one, and the reply-rate cost is already in your numbers if your sequences read like everyone’s. I tell every team the same thing: AI is the prep work, humans close the deal (and this requires diligence from your side, I know this myself). Use it to build a real point-of-view on the person, the team, and the market, then send one specific human line, not another blast. And stop reverse-engineering the 498 emails that failed. Look at the 2 that booked and build from the win. On a lean team across markets, one specific first line beats volume every time.
The European read: Multi-language outbound makes this sharper. A generic opener translated into a buyer’s second language reads flatter and more machine-made, not less. The account-specific detail, named in their market, is what makes a lean team’s outbound land where a big team’s volume bounces.
Do first: Open your live sequence and read the first line of step 1 out loud. If it could have gone to any account on your list, it is generic. Mark it to rewrite.
Do this week: Rewrite that first touch so the opening sentence names 1 specific detail a competitor’s blast could not contain. Give your AI 1 prompt that pulls a real trigger per account, funding, hire, launch, product change, and drafts the opener from it, then strip the AI tells before it sends.
Do this month: Build the account-detail-plus-voice step into your standard outbound motion, so specific and human is the default first touch, not the exception a top rep makes by hand.
The rule: If the first line could have gone to any account on your list, it is spam with better grammar. Name 1 thing only that buyer would know.
You know it worked when: Reply rates on the reworked sequence climb and prospects answer the specific line, not the ask.
The play: 5 steps to increase your outbound meeting-booked and held rate (2026 series). Rebuilds the first touch around one specific, human detail per account so more meetings get booked and held. Best for 2-5M teams scaling outbound across markets.
Sources: Ruben Hassid, “Can you detect AI?” (July 22) · Charlie Hills, “I made myself replaceable (and hired 3 people)” (July 19)
Problem 3: Your reps still do your highest-value deal work by hand, and it is eating your senior selling time
There is deal work your best reps still do by hand on every deal, call prep, the mutual action plan, the multi-thread research, the tailored proposal, because the last time you pointed AI at it the draft was too shallow to send. That verdict has a shelf life, and it just expired. AI can now carry the research and the first draft of those deal artifacts in a form your rep would actually use, in each market’s language, which hands your senior people back the hours the deal is actually won in. On a lean team spread across markets, reclaimed senior selling time is the difference between covering your pipeline and dropping deals you already paid to source.
My take: Founders shelve a task once, mark it not-AI-ready, and never look again, then the frontier moves and the verdict is stale. But the real lever is not the model, it is the context. I had my team analyze every closed-won and closed-lost deal from a year and pull the pattern that repeats: the pain, the objection that kills, the line that wins. Loaded with that, the AI draft reads like our best rep, not the internet. AI does the first draft. A human does the second and sends it. The point is the deal moved, not which tool wrote it.
The European read: A lean team spread across markets cannot put a specialist behind every deal. AI carrying the multi-thread research and the first draft of the deal artifacts, in each market’s language, is how your senior reps spend their hours where the deal is won instead of on prep.
Do first: List the 3 deal-cycle tasks your reps still do fully by hand, and mark the 1 that eats the most senior selling time.
Do this week: Rebuild that 1 task as an AI-drafted first pass loaded with a winning deal’s context, and compare the draft to what your best rep would send.
Do this month: Set a quarterly re-audit of what your deal cycle keeps manual, so a shelved task gets retested as the tools improve, not left parked for a year.
The rule: Not-AI-ready has a shelf life. Retest a shelved deal task each quarter, and judge it on whether the deal moved, not on which tool drafted it.
You know it worked when: One deal task moves off your reps’ plates this quarter, and your senior selling time goes up without a quality complaint.
The play: 5 steps to make your AI deal drafts sound like your best rep (2026 series). Loads your winning-deal context so AI-drafted deal work advances the deal instead of reading generic. Best for 2-5M and 5-10M teams shortening the deal cycle.
Sources: Ruben Dominguez, “Anthropic just cut the price of frontier intelligence in half” (July 24) · Ruben Dominguez, “China just open-sourced Opus-level intelligence” (July 19)
Save this. 3 AI moves from this week’s Claude Signal, each on a real revenue motion, with my operator take and a move you can run today.
Send it to 1 founder or GTM operator scaling toward 10M in Europe who reads everything and ships none of it.
Steal this move: Build a voice-loaded content system that scales past you in 4 weeks
Problem 1 is the most urgent, because demand is where most lean teams already turned the AI volume up. Here is the hero play in full.
Play: 5 steps to make Claude write in your voice, “5 steps to make Claude write in your voice (2026 series)”
Fixes: A content engine that scaled on AI and started reading like generic, flaggable AI instead of like you.
Best for: 0-5M founder-led demand, and the European reality of building a recognizable voice in each market instead of shipping the same translated generic content everywhere.
The 5 steps (Define / Build / Validate / Operate, over 4 weeks):
Week 1, Define: Pick 1 recurring content type. Pull your 5 best past examples. Write the voice file: the 5 lines that make your best work sound like you, plus the AI tells to strip.
Week 2, Build: Draft against the voice file, then score each draft against a rubric built from your best and worst past pieces. Add the 2 gates, does this sound like us and would a detector flag it, and 1 human approval before publish.
Week 3, Validate: Run it for a week against your manual output. Compare engagement or reply rate, and tighten the voice file where it drifted. Route the drafting to a cheaper model, keep the top tier for the rubric, and confirm quality held.
Week 4, Operate: Add a scored idea queue so effort goes to what performs, and hand staging or publishing to someone with you as the final gate.
Standing: Feed every miss you catch back into the rubric, so the system sharpens each week instead of drifting generic.
Template: A 1-5 voice rubric. Rows: sounds like us, names a specific detail, no stock opener, no tidy 3-part summary, would pass a detector. Score each draft 1-5. Anything under 4 goes back through the voice file before it ships.
Paste this into your AI:
Here are my 5 best-performing posts and my 3 worst: [paste]. Extract the 5 lines that make my best work sound like me, and the AI tells to strip. Then draft my next post on [topic] against that voice file, and score your own draft 1-5 on: sounds like me, specific, no stock opener, no tidy summary, detector-safe.
Full play, template and workbook inside GTMcraft OS. Reply or DM me to get access.
Also on the radar
Ruben Dominguez, “China just open-sourced Opus-level intelligence” (July 19). An open model tied the top score and undercuts on cost. Your fallback and cost options for GTM work just widened. Price 1 high-volume GTM task on an open-weight model against your current default.
Ruben Dominguez, “Anthropic just cut the price of frontier intelligence in half” (July 24). The new top model ships an effort dial. Running every GTM workflow on maximum overpays. Turn the effort down on 1 routine GTM task and check the quality holds.
Ruben Dominguez, “Anthropic just cut the price of frontier intelligence in half” (July 24). The quality-per-dollar jump lands on retention too. Draft your next renewal or QBR on the new model with the account’s usage and wins loaded in first.
Why this matters now
Every team you compete with now runs the same competent AI, so producing more output stops being an edge the moment everyone can. What does not commoditize is the layer only you own, your voice, your account knowledge, and your read on what good looks like in your market. This builds on Embed The Edge from July, where the move was wiring AI into your revenue motions. This week is the next beat: make sure what comes out the other end still sounds like you, in every market you sell into, so it earns the reply instead of the skim.
What are you seeing this week: is your AI content or outbound starting to read generic, and how are you catching it before it ships?
Reply or send me a DM.
Go deeper with GTMcraft OS
Scale on plays, not prompts.
You do not need another newsletter, podcast or LinkedIn post. You need the system under it.
Read the plays. Every single play is free and live in GTMcraft OS on Notion. Click through from any edition.
Build your own. GTMcraft OS members connect the full libraries, 100+ plays plus 100,000+ LinkedIn, newsletter and podcast signals, straight into their own AI or Claude context, so their team builds on the whole corpus instead of one-off prompts.
The single plays are free to read. Wiring the whole corpus into your own operating system, and keeping it current, is what the membership gives you.
Curious to see inside GTMcraft OS, DM me or reply on this email.
Master your GTM skills AI can’t replace inside GTMcraft,
Koen
Your (human) GTM Agent
PS.
