How I Used AI and Hiring to Move Out of Email Agency Fulfillment

I'm Nick Yates, and I run a boutique email marketing agency for e-commerce brands. My aim was never a huge client roster. I wanted higher retainers and a smaller operation, but I had become the person doing fulfillment all day.

When Daniel introduced this interview, he put the agency at around $40,000 a month, with an additional roughly $20,000 a month in revenue over about three months. The practical change I described was moving into an operator role through hiring and AI. Those revenue figures were not a profit calculation or a confirmed measure of contracted MRR.

I had built a business that depended on my own delivery

I started as a copywriter, moved through direct response work and found my way into email marketing and retention. Even after years of practice, a fast email could still take me about 30 minutes. Coming up with a range of fresh angles for each client was a different problem from simply writing quickly.

I described fulfillment as my bottleneck. It took up my day before the Slack messages and emergencies were even accounted for. The agency could not grow well while every piece of client work had to come through me.

Client Ascension related reading: Steven's content agency case study offers another service business's account of delegation. It is a separate example of founder capacity, not evidence of my results.

Losing retainers exposed the risk in a boutique model

After my first year in Client Ascension, I stepped away. Within about 90 days, I lost around 80% of recurring revenue and eventually got below $10,000 a month. With larger retainers, losing a few clients was a substantial blow.

I returned almost a year after leaving. I missed the accountability and having coaches to ask about offers, sales, operations and mindset. The AI side was the tipping point for returning, but I also had to move past trying to solve everything on my own.

The swipe-file agent gave the writing system references

Repetitive output was a problem with AI-generated emails. I built an agent around a large swipe file of emails from leading brands. Before producing an output, it had to reference that material and look for an appropriate fit for the email we were writing.

That library was not automatically filling itself from monitored brands. When Daniel asked about that, I said it was something I wanted to reach, not a feature already working. The useful system I had built still depended on the examples available to it.

Shared instructions replaced a collection of separate prompts

Earlier, each client had a strategist and copywriter workflow in Claude projects. With seven or eight clients, a change could mean updating roughly 16–20 prompts and then telling the team to update their own Claude setup.

I moved the system into Claude Code with a shared GitHub repository. My team and I could work in the same system, with global changes instead of repeating the same edit across every prompt. I gave a quick instruction change as an example of the difference, rather than a benchmark for all maintenance work.

Hiring changed what I did with the day

I went through the hiring process with guidance on the roles, rates and margins to work toward. The system I had built could then be handed to an employee working with AI, rather than leaving me responsible for writing every email myself.

I said I was doing pretty much no fulfillment by the time of the interview, but I still monitored Slack, looked at designs in Figma and improved the agents when I saw an issue. My work had shifted toward helping the team and outreach; I had not stopped overseeing the business.

Client Ascension related reading: the acquisition guide discusses the sales side of an agency. It is useful alongside this delivery story because freeing capacity and building an acquisition process are different jobs.

The next constraint was outreach, not another email draft

I was still learning how to use the extra time. My instinct was to go back into reviewing emails and designs, but I was trying to identify the current constraint instead. At that point I was focusing on Dream 100 outreach and generating more sales calls.

My growth to that point had come through good work, referrals and an organic presence. Cold outreach was an opportunity I wanted to develop, not the proven source of all the growth described here. Later revenue targets were goals rather than results I had already achieved.

Client Ascension related reading: the AI Assisted Agency program page has current program details. My account describes my experience at the time, not guaranteed results or a statement of today's inclusions.

Adapted by Client Ascension with AI assistance from Nick Yates's interview. The first-person narrative follows the source speaker’s account. Watch the original video