The short answer
Use AI to sharpen what you've already written, not to write it for you. Give it your own material first (your target list, your scripts, your real case stories, your meeting notes) and ask it to refine, adapt or check. When a chatbot writes prospecting from nothing, it invents results and sounds like everyone else's inbox.
The problem with "write me a prospecting email"
Ask a chatbot for a prospecting email with nothing in front of it and you get two problems. It sounds generic, like the fifty other AI emails your prospect deleted this week. Worse, it makes things up: close rates, client results and savings numbers that read as plausible and are completely invented. In insurance, a producer who sends an invented result to a prospect has a real problem.
The rule: refine, don't generate
AI is excellent at making good material better. It's poor at creating good material from nothing. So the order matters:
- Start with your own material. Your target accounts, your scripts, your real case stories, your notes from the meeting.
- Ask AI to do one specific job with it. Tighten this script. Adapt this case story to a construction company. Pull the key points out of these meeting notes.
- Check before anything goes out. Every claim, number and promise gets checked against what's true and what your agency can stand behind.
Five places AI actually helps a producer
- Pressure-testing the target. Paste your ideal client profile and your last ten wins and ask where they don't match.
- Tightening scripts. Paste a script in your own voice and ask for a shorter version that keeps your phrases.
- Adapting follow-up after a meeting. Paste your notes and have it pull out what the prospect actually said, quoting their words and flagging anything it isn't sure about, before it drafts a single email.
- A standards check. Before a message goes out, ask it to flag any claim, promise or number you can't back up.
- Research and prep. Summarize what's public about a company before a call, so the time goes to the conversation instead of the digging.
How to keep your voice
Tell it who you are and how you talk, and give it examples of messages you've actually sent. Ask it to keep your phrases. Then read the result out loud. If it doesn't sound like you on a phone call, it isn't done.
What AI shouldn't do
It shouldn't invent a case story, a result or a close rate. It shouldn't pick your target for you from nothing. And it shouldn't replace the relationship. AI removes the guesswork and the prep time. The trust is still built by a person.
How we use it at SOS
We call this the Force Multiplier. Every producer gets a set of AI prompts that only work once they paste in their own Personal Runbook, so the AI is refining their real target, their real scripts and their real case stories instead of inventing new ones. Producers spend less time on research and prep, and more time building trust.