You're still rewriting every draft the machine hands you. Here's one that writes like your house and learns from your red pen.
You know the drill. The first draft comes back fast. It reads like everyone else's. Your writers clean it up, prompt again, clean it up again, and you still read every line. Somewhere in there somebody is copy-pasting between six chats to get one ad out.
Here's what's underneath it. Your best work isn't anywhere the machine can see it. Every promo that beat a control, every note you've written in a margin, sits in a folder while the chat starts from zero. And when you try to fix that by stacking everything in, it gets dumber, not smarter.
I build a system in your house that writes from your winners, in your voice, and drafts in passes the way a writer does, so by the time it's polishing, the only thing in front of it is your best copy. Then it learns from your red pen. What you fix this week, it gets right next week.
The junior work around a VSL, the ads, the landers, the upsells, the checkout copy, gets done without a junior.
I built it for my own funnels first. Now I install it for copy teams and teach your chief to run it.
It isn't the machine. It's what the machine can see.
It drafts from the internet's average.
Every winning promo your house ever ran, every reason a control beat the last one, every note in your margins: none of it lives where the model can see it. So it writes from what it can see, which is everyone else's copy. It comes back sounding like the internet, and the rewrite starts.
So "using AI" means a slower intern.
Writers prompt, clean up and prompt again. You still read every line. The work didn't shrink; it moved. Meanwhile the ask from upstairs is more packages, no new hires, and a straight answer on why the turnaround hasn't dropped.
The fixes you've been handed don't fix it.
Your tech-savvy friend told you to dump the swipe file into a project. You did, and it got dumber, because now every chat carries everything. An engineer can wire the tools, but an engineer can't tell a strong lead from a weak one or pick which control to build from. The system has to be designed by someone who writes copy that sells, then handed to your team in a shape they can run.
Nothing your team learns this week is in next week's chat.
You fix the same thing every week. The machine never finds out. Your writers get faster; the drafts don't get better. That's the gap nobody's named to you, and it's the one that decides whether this ever pays for itself.
Your winners, your voice, your frameworks. Set up so the machine writes from proven bones.
"I'm never going to build a promo from scratch. I'm always going to swing from solid footing, and that means building from proven bones." That's how my own copy gets made. The install puts it in your house, on your material, in the order it pays off.
-
1
Your winners are the source.
Your controls, your offer data, your house frameworks and your voice, organized so every draft starts from what has already sold to your list. Not a blank prompt, and not the swipe file dumped in whole.
-
2
A copywriter's craft, and it drafts in passes.
How to write a lead, an ad, an upsell, a checkout page, by asset and by market, built in from eight years of writing front-end funnels. It works the way a writer does: decide what the section has to do, pick the proven structure, write it, polish it, check it. Each pass sees only what it needs, so by the time it's polishing, the only thing in front of it is your best copy. The calls you've made stay made; nothing gets relitigated.
-
3
A loop per job. The ad loop is the showpiece.
One big idea and one VSL become a stream of ad packages you'd actually test. Same for the lander, the upsells, the checkout. This is where volume comes from, and it's where every chief I've walked through it leans in.
-
4
It learns from your red pen.
You edit it the way you'd edit a writer, and it takes the note. Your edits and your calls get captured and folded back in, so what you fix this week it gets right next week, and the pile on your desk shrinks because the evidence says it can. A tool subscription can't do this. This is the part that compounds.
-
5
It checks its own work before you see it.
Compliance, made-up numbers, length. A fail goes back to be redone; it doesn't get patched over and passed along. You review the finished draft, not every line.
What stays human: the big idea and the front-end VSL. Too much taste in there to hand off. Everything around it gets handled.
Two ways in. For companies: a done-for-you build plus a retainer to bring your chief and team up to speed. For copywriters: the build plus coaching from a working DR writer. Same machinery, sized to the seat.
Chiefs with winners on file and more to ship than hands to ship it.
A good fit
- Copy chiefs and creative directors at publishers and D2C brands running paid traffic at real volume.
- A body of controls to build from, and a chief who'd rather run the machine than clean up after it.
- More packages due out the door than the team can write, and a hiring freeze upstairs.
- No engineer on staff and no plan to become one. I build it, install it, teach you, then get out of the way.
Not a fit
- No winning copy on file. There's nothing to build from, and I won't pretend otherwise.
- Looking for a tool subscription. This is a build and a working relationship, not a login.
- Wanting the machine to ideate the big idea for you. That part stays with a human, and it should.
Apply. I build a demo on your copy. You watch it run.
No deck, no generic walkthrough. Tell me the one copy job you'd most want off your desk. I build a working demo around it and show you live: what goes in, what comes out, and how your chief would run it.
Apply
Six fields, most of them pick-lists. Two minutes.
I build
On your actual situation, not a canned example. If I can't build something worth your time, I'll tell you that instead.
You see it live
We get on a call. You watch it work, ask what you want, and decide whether it earns the next conversation. Free either way.
Robert Bennett, direct response copywriter.
I write front-end customer acquisition funnels that scale on cold traffic: Facebook ad to lander to long-form VSL to the day-zero upsell flow. Specialist, not generalist. Eight years in the income opportunity market, with growth opportunity as the second lane.
Two years ago my desk looked like yours.
"Before this, I ran my whole business out of Google Drive: client folders, project folders, dashboard logins for client data, heaps of research, swipes, dozens of prompts I'd version in isolated chats. Tons of copying and pasting, tons of shit output, frustration, exhaustion, wondering if I was wasting my time and would be better off writing the old fashioned way, manually (ripping my hair out over line 17 on page 20, writing and rewriting ads, drafting checkout pages by hand)."
I tried projects. Worse, because everything got stacked on every chat. I tried an autonomous agent. A disaster: too autonomous to trust, no memory to speak of, and the copy got worse. Then I put the whole business in one place, gave each job only the context that job needs, set up every drafting session before it starts, drafted in passes, and built a real feedback loop off my own edits. That's where the wins started compounding.
"Once I draft a front-end VSL, everything around that, like the ads and landers and upsells and checkout copy, is so easy that I hardly have to be in the loop at all."
I built this to survive my own client load. Someone was going to package it for copy teams, and I'd rather it be a copywriter than a software company. "Yeah, I'm selling my advantage, but someone's going to sell this eventually whether it's me or not, and I'd rather ride the wave than be consumed by it."
- Front-end funnels on cold traffic. Ads, landers, long-form VSLs and upsell flows, built to scale on paid.
- Eight years in one market. Income opportunity primary, growth opportunity secondary.
- Runs on his own work first. The same system writes the copy that goes out under Bennett DRM's own name.
- Chiefs the machine like a writer. Draft from proven comps, edit in passes, test four to five big-idea variations.
Tell me what you sell and what you want off your desk.
That's enough for me to build the demo. You'll book the call on the next page.