Case study · AI delegation · Live demo

20 emails · 6 sequences · 9 automation rules, rebuilt in one working session. Full story below.

The campaign belonged to Wonder Wardrobe, a fashion-education brand we work with. The plan: 20 emails across 6 automated sequences, plus the offer, the audience rules, and the landing page copy. All of it lived in one Google Doc that had grown to 23 pages.

In the comments, the art director was asking which image went where. The marketing director was trying to work out which stage of the campaign she was even looking at. The copywriter kept answering both of them with pointers to headings. Scrolling 23 pages had quietly become a project-management method. The launch was getting closer, but the campaign was nowhere close to being published.

If you have ever run a launch out of a doc like this, none of that will surprise you. 

In one working session, the Delegatos team rebuilt the document with AI into an interactive command centre. Every email, sequence, and automation rule on one screen, with 6 hidden errors flagged along the way that nobody had caught by reading.

This case study shows how a team can use AI to organise and streamline any project: the before, the after, everything the rebuild caught, and what we delegated next, with a live demo you can click. 

“The doc wasn’t wrong. For writing copy, one long page made sense. But the art director couldn’t see which visuals went where, the marketing director couldn’t follow what each stage was for, and the gaps got papered over with endless tagging and comments. The command center gives all three roles the same screen.”

— Daria Andronescu, founder, Delegatos

We must admit, the copywriter wrote an amazing email marketing campaign. But the way he arranged it in a doc makes perfect sense only to the person who wrote it and to nobody else on the team. 

1. Structured for the writer, not the team. For writing copy, one long page with headings made sense. The art director read it and couldn’t tell which visuals went where, or why. The marketing director couldn’t follow what each stage was for. So everyone tagged everyone. The comment threads became the real chaos for everyone.
2. Automation rules buried in headings. The logic that decides who receives what lives in section titles like “Automation: Get a Yes reply to send to Conversion Email 1.” The problem is that reading does not reveal the campaign’s real structure and logic. There was no way to hold the full picture at a glance.
3. Duplicated copy that drifts. Those two versions, nine pages apart? Conversion Email 4 contained the entire body of Email 1, pasted in as a “forwarded message.” 
4. Broken links you can’t see. In a Google Doc, a link’s visible text and its actual URL are two different things. The offer page link read one URL and pointed to another. A testimonial email sent readers to the wrong article.
5. Placeholders that never resolve. “Countdown Timer: TBA.” “Insert counter – code.” “Here is your account activation link: ____.” Thirteen image slots marked with asterisks. A doc has no way to make unfinished work visible, so it stays unfinished.

None of this is carelessness. It is what happens when if-then paths are kept as pages of text. 

McKinsey Global Institute estimated back in 2012 that knowledge workers spend roughly a fifth of the workweek just looking for information. A 23-page campaign doc is your own private version of that tax.

Picture the whole launch laid out on one screen, like a map. Every email is visible. Every rule is written on the line between the emails it connects. Every gap is marked.

We took the doc it exactly as it was, comments, TBAs, and blanks included. Tidying would have hidden the problems we wanted caught, and frankly, nobody had the appetite for one more pass through 23 pages. Our tool for the build was Claude, which hosts the result as a shareable interactive HTML page; other capable AI assistants can do the same job. 

One working session later, the plan had a new shape:

• Overview: the offer and its math (the €79 class, three bonuses worth exactly €79, the 7-day deadline), the audience source, and the exclusion lists.
• Campaign flow: all six sequences as a visual map. Each email is a clickable tile that opens its full copy. The automation rules sit on the arrows between the tiles, where logic belongs.
• Flagged issues: everything wrong or unfinished, ranked by severity, each linked to the email it lives in.

Twenty emails and nine automation rules, all visible at once. That view simply did not exist before. The closest thing the team had was scrolling.

The real change is what the week feels like for the five people running the launch. The campaign became something the team works on together instead of something each person reads alone. That happened because the command centre works like a small app you click around, not like a document you read top to bottom. 

Five features carry the difference.

Preparing the launch was half the job. Once every email was marked ready, we connected Claude to Wonder Wardrobe’s email marketing platform (via MCP, a standard that lets AI work inside other tools) and tasked it with the build itself: all 20 emails and all 6 automations, created directly in the platform. 

The entire build took a few hours. Done by hand, copy-pasting each email and clicking together each automation, it would have cost the team about two-three more days.

This is AI delegation in practice: the team decides what good looks like, and the AI does the routine build.

That build deserves its own post; subscribe so you don’t miss it.

Product names, prices, and URLs in the live demo are altered; every defect described is from the real document.

Daria Andronescu

Founder, Delegatos

AI can’t do the work that actually matters: the thinking, the judgment, the ideas. So I delegate the repetitive work for both of my companies to it, and I help other founders do the same. Connect on LinkedIn →

Most growing teams lose them the same way: scrolling, tagging, waiting for someone to confirm which version is current. We turn AI tools into working systems, so your people spend their time on judgement and ideas instead. Send us the doc as it is. The mess is exactly what the rebuild catches.