The Skepticism
Our first instinct was "AI can't write like we do. Our clients know our voice." Fair concern. Clients work with us partly because of how we communicate.
But our team was spending 30+ hours a month writing reports. At $75/hour, that's $36,000 per year on routine reporting.
So we asked: What if we let the AI write it, and our strategist edits it? Would that save time without sacrificing quality?
The Experiment
We built an agent that writes our monthly performance reports. It:
1. Pulls performance data from Google Ads, Meta, LinkedIn
2. Compares to prior month and prior year
3. Flags what's working and what isn't
4. Writes the report in our format and tone
5. Creates recommendations for next month
The first draft took 30 minutes of AI processing time. Our strategist then spent 60 minutes editing, rewriting sections, adding client context.
Total time per report: 90 minutes for final, client-ready output.
Before: 150 minutes to write from scratch.
After: 90 minutes to edit from draft.
That's 40% time savings.
The Surprising Part
The AI writes good reports. Not perfect. Not award-winning. But good. Thorough. Organized. Clear recommendations.
What's interesting is that clients sometimes prefer the AI version because it's more structured and consistent. Some of our human writers have... stylistic quirks. The AI doesn't.
Clients still know it's from our team. The voice is ours, not the AI's. The strategic thinking is ours. The client context is ours. The AI is just doing the assembly.
What Needed Fixing
Early versions sounded slightly robotic. We adjusted the prompt to be more conversational. Fixed.
The AI sometimes missed client-specific context. "This was a good month because of the seasonal factor we discussed in kickoff." The AI didn't know the client context. We added a briefing step where our strategist provides context before the AI writes.
Recommendations were generic sometimes. "You should optimize your bidding." That's true but not useful. We built in more specificity requirements to the prompt.
The Decision We Made
This is now standard for us. AI drafts the report. Strategist reviews and edits. Client gets a better product faster. Win.
Not everyone loves this. Some people think "AI writing client-facing content" is Scary. But clients like the reports. Quality is consistent. Process is faster. That's what matters.
The Pattern Here
This is different from total automation. The AI didn't replace the strategist. It replaced the drafting phase. Strategist still owns the final output.
That's what works: AI for the routine work, humans for the judgment calls. Not AI replacing humans. Humans and AI working together.
What This Means for You
If you have people writing routine client communication — reports, summaries, status updates — this pattern works. Build an agent to draft it. Have someone review and customize. Ship it.
Your team does the work that matters. The AI does the assembly. Clients get better communication. Everybody wins.