What I tried
I wanted to be a guest on more podcasts to talk about how small teams put AI to work. Pitching one show at a time never got done, so I built a pipeline that worked from a list of 137 business podcasts. It read what each show covers, wrote an episode idea for that show, drafted a short pitch in my voice, and checked each draft against rules I wrote. The first background job to rewrite the pitches finished but produced nothing because the worker could not read the files it needed. The earlier drafts buried the ask under credentials and used long tracking links that looked like spam. I changed the rules: put the ask on its own line after the opening, use one sentence of credentials and one plain link, and write no more than three sentences before the ask.
What happened
On Oct 9 2026, I sent 137 pitches in one day, 130 by email over a few hours and 7 through show web forms, with 0 send errors. Every draft passed the rule check, and nothing was sent until I approved the wording. About three hours after the batch ended, 14 shows had answered: 2 hosts sent booking links, 4 said no, 1 asked a question, 5 emails bounced, and 2 were auto replies. One recording is now booked for Oct 16. I sent the seven replies myself after reading each one. The writing was not the hard part. The rules were. Once the rules said exactly what a good pitch looks like, 137 pitches were easy. I also learned to check the output of a job that says it finished. What would you pitch if sending 137 of them cost you one afternoon of writing rules?
Comments · 1
How did you source them and find their contact info? Is there a type of podcast you were looking for, especially in terms of size of podcast?