“AI is eating part of
what my company
sells. I’m not going
to pretend
otherwise.“I run a software development shop. Some of what shops like mine sell is getting cheaper and easier to replace every month, including some of what we sell. I’m not going to stand in a room and claim different.
Here’s what I’ve watched happen instead. The building got easy. The keeping got harder.
Someone still has to own the software after it works. Someone still takes the 2 am call when it breaks with a client’s data inside it. That part hasn’t moved at all.
Most of what gets built at events like this one stops working within ninety days. Nobody owns it, nobody documented it, and the person who built it got busy. So the last stretch of the afternoon is about those ninety days rather than the build.
You’ll leave with software that runs. I’d rather you leave with software that’s still running when you’ve forgotten this Wednesday happened.
Brian Gerstner White Label IQ
One layer
that makes every tool
know your agency
An AI workshop for agency owners, built and run with Shahed Islam, who runs AI build sessions in New York.
What you build
A control tower. One layer that holds what the AI needs to know about your agency and sits above every tool you use:
- How you actually run a project, not how the SOP says you do
- Your standards, and examples of work you’ve graded with the reason attached
- Your client history, and what’s already been decided
- What’s off-limits, because a wrong document is worse than a missing one
The part that gets skipped:
The building is the easy half. What kills these things is that nobody owns them and nobody’s looking. That’s the last stretch of the afternoon.
What you need
A laptop, and one real problem in your agency you’d pay to make go away. Nobody writes code by hand.
3 reasons to skip it
(we mean all of them)
You’ve never used
ChatGPT or anything like it.
You’ll be lost by minute ten, and we won’t slow the room down.
You already run AI in production
with client data going through it.
You should be teaching this workshop.
You want to sit
and watch a keynote.
Wrong Wednesday. You’ll be working for three hours.
Now, the one reason to come.
You have a stalled AI project you’d like to finish.
The tools weren’t the problem, and they won’t be next time either. What decides whether this works is how you think about the thing underneath, and that’s what the afternoon teaches. Learn it once and it holds through every tool you use after this one.
You think you’re the only one stuck. You aren’t. 84% of agency leaders say they’re adapting their operations for AI. Between 13% and 16% have finished the thing they called most important themselves. Nobody in your peer group has said it out loud.
It’s the most common condition in this industry, and it’s fixable in one afternoon.
Build it on Wednesday.
Start using it on Thursday.
The control tower
Shahed built it, and he’ll build it with the room. It’s a management layer for your agency’s context, sitting outside every other tool you use, so what the AI knows lives in one place you control instead of trapped inside whichever product you happened to start with.
You’ll build yours on a real problem from your own agency, on seed data we provide. Nothing of yours connects to anything, and you leave with it running.
And the structures behind it
Governance, gates, loops, and the system prompts you can use on Thursday morning. This is the part that decides whether the thing still works in January, and it’s the reason the afternoon doesn’t end at the build.
84% say they’re adapting.
Between 13% and 16% have actually
shipped.
So the pressure is rising and the execution isn’t, which means the gap between saying and shipping is the widest it will be. It closes as the rest of the field catches up.
Their clients aren’t waiting.
They still want your thinking.
They’re buying less of your
doing.
Read it at agencycore.org
These are from Agency Core and Agency Edge, two primary research studies of 579 agency leaders and 400 of their clients. Primary research on this industry barely exists, so most agencies are working from guesses. I co-founded Agency Core, so weigh it accordingly. It’s public, the methodology is on the site, and you don’t have to give anyone your email.
Brian Gerstner – White Label IQ
What happens
in the room?
Systems, not prompts
Why the tools you’ve tried didn’t stick, worked through against your own agency rather than a made-up example.
The build
Your control tower, on a real problem from your agency, working before you leave.
What keeps it alive
Who owns it by name, what data it’s allowed to touch, and what happens the first time it’s confidently wrong in front of a client.
Food and drinks
Same room, no agenda, nobody presenting. Format and details in your confirmation.
Full agenda published shortly.
“I’ve built
one of these before.
It’s not running now.“
That’s the usual ending, and it isn’t because the thing was built badly.
It went home with nobody’s name on it. No test that would catch it going wrong. No ten minutes a week to look at it. So the person who built it got busy in week three, and it sat there.
Three hours plus food and drinks. Bring one real problem.
We’d rather you didn’t need us twice. So we’ll show you the thinking behind it: how to put it together so it holds, how to pick what’s worth building next, how to spot it going wrong before a client does.
You don’t leave with our system.
You leave able to build
your own.
And you’re not doing it alone in there. Brian, Shahed and Erik are in the room all afternoon, waiting to be useful. Pull one of them over, show them what you’re stuck on, and get a straight answer.
Who’s
running it?
Meet people behind the system.
Three people are asking for your Wednesday afternoon.
Worth
checking all three before you say yes.
Brian Gerstner
White Label IQ
Builds and runs software for agencies, with their clients’ data inside it. Which means when it breaks at 2 am, his phone is the one that rings. He also co-founded the Agency Core research on this page.
Shahed Islam
ManagedCoder
Started SJ Innovation in 2004 and somehow ended up with 140 people in four countries. Built Control Tower because he needed it, then discovered everyone else did too.
Erik Martinez
Director of Client Strategy, White Label IQ
Twenty-five years agency-side, including Blue Tangerine and Digital Velocity. He has heard every version of “we’re going to build that internally” and knows how those usually end.
Why
it’s invite-only,
and why there’s no ticket
An afternoon in a room of serious agency owners is worth more to us than ticket revenue. Charging would change who walks in, and so would opening it to anyone who clicked.
So we invite, and we cap the room. Nobody is pitching from the stage and there’s no offer waiting afterward. If you’ve landed here from something else we do, the same applies. If you want to talk about working together, find us at dinner.
Request
an invite
One Wednesday afternoon in Manhattan, and you go home with a
working control tower for your agency and the structures to build the next one.
The room is by
invitation, and it’s capped. Requests close Tuesday, October 27.
* indicates required fields
Ai Workshop for Agencies
"*" indicates required fields
What people ask before they say yes
I bet I need to be technical.
You don’t. Nobody in the room writes code by hand.
How much do I need to know about AI?
Enough to have used one of these tools for real work. Past that, the room holds owners of twenty-person shops alongside people who’ve already shipped their own internal tools. Both will be busy for three hours.
Three hours isn’t long enough to build anything real.
It’s long enough to build something small that runs, which is the only kind worth starting with. Ambitious first projects are one of the two most common reasons agency AI stalls. You’ll build something narrow on purpose.
I don’t have a full afternoon.
Then skip it rather than half-attend. Three hours, then food and drinks if you want them.
What do I bring?
A laptop, and the AI project you already started and never finished.
Where exactly is it?
Manhattan. We send the address, plus anything you need set up beforehand, well before the day.
Does my client data go anywhere?
No. You build on seed data we provide, on a real problem from your agency. Nothing of yours connects to anything. The layer you design is the thing you take home, and you point it at your own data when you’re ready and it’s your call.
Can I send someone from my team?
Yes, if they can make decisions about how your agency works.
