What we watched, and the order we fixed it in.
Most companies arrive at AI email from the outside, as a feature to bolt on to something that already exists. We arrived from inside the thing being replaced. George spent six years running SmartrMail, and the conclusion did not come from a market report. It came from watching his own paying customers use the product he had built for them.
"I built SmartrMail (12K customers, billions of emails, acquired 2022) and watched every single customer waste hours dragging composer blocks doing work that AI can now do in seconds. The gap between what AI can do and what email platforms let you do with it is enormous."
That is the observation this company rests on, and it is not a complaint about dashboards looking dated. It is a specific charge. The work people were doing by hand in a composer was work a model could already do, and the interface was the only reason they were still doing it by hand.
We fixed the knowledge problem before we shipped the product.
This is the part of the story that runs against our own sales pitch, which is why we lead with it. In February 2026, five months before Nitrosend launched, George published the Email Marketing Bible: a free, open source reference now built from 908 sources and the published insights of 49 practitioners. He published it because AI is bad at email.
"It's 2 problems: the tools don't connect well to AI yet, and AI doesn't actually know email marketing nuance. Ask it to write a welcome sequence and you get generic, unsourced fluff. Ask about deliverability and it confidently makes things up."
Saying that out loud while selling AI email is uncomfortable, and it is still true. His response was to take the two problems in the order he could actually solve them. "I couldn't fix problem one in a week," he wrote, so he fixed problem two first and gave the answer away rather than selling it. The rule that came out of that work is the one we build against now: "The fix for mediocre AI-generated email isn't better prompts. It's better references."
It is why Nitrosend ships a Brand Kit and a separate AI Memory document instead of a longer prompt box. The agent that writes your email reads your colors, fonts, logo, voice and stated strategy before it writes a word, because a model with no references produces slop in any tool, including ours.
Then we removed the dashboard.
The platform came second, and by then the argument had already been made in public:
"The dashboard was never the product. It was the bottleneck."
The version of that claim we would defend in a room is narrower and more useful. Every campaign anyone has ever dragged and dropped was, underneath, a single sentence of intent that a person then spent an afternoon translating into blocks. The dashboard was the translation layer. Once a model can do the translating, the layer is cost, not product. George's own framing of the shift: "In 2026 the job changed. Agents now build the campaign, segment the audience, draft the copy, and stage the send. The marketer is the director, not the operator."
That is a real change in what the job is, not a change in who clicks the button. A director sets intent, judges the result and approves it. An operator assembles it. We think the assembly is over, and we are not neutral about it, because we built two companies on the assembly.
If you have questions, want to argue with any of this, or just want to say hello, reach us at [email protected].
