Tech Training
How I'm Using AI to Write Better Show Notes (Without Losing My Voice)
By Nina Capone · May 27, 2026Weekly · Editorial Assist

Let me be real with you: I was skeptical as hell about AI until I hit a wall. Running In Da Streets Radio means I'm recording 3-4 episodes a week, and writing show notes after a 90-minute conversation was eating my whole next day. I needed help, but I wasn't about to let some robot sound like me. Here's what I figured out.
My current setup: I record interviews in Riverside (which auto-transcribes), then I feed that raw transcript into ChatGPT Plus ($20/month). I use a custom prompt I built that says 'You're writing for a Philly indie radio audience, keep it conversational, pull out the 3 best quotes, give me 5 timestamp moments, and write a 150-word episode description.' Takes about 90 seconds. The output is maybe 70% there.
That last 30% is critical — that's where I go back in and add my voice. AI loves to say stuff like 'delve into' or 'landscape' and I'm like, nobody in Philly says that. I swap it for 'we got into' or 'the scene'. I also check every single quote because sometimes the transcription mishears slang or names. You cannot just copy-paste. You'll sound like a press release.
For social media posts, I'll ask ChatGPT to give me 5 different caption angles from the same episode — one for Instagram (casual), one for LinkedIn (professional), one for Threads (spicy), etc. Then I pick the one that feels most like something I'd actually say and tweak it. This saves me from staring at a blank text box for 20 minutes. I'm using the tool to beat writer's block, not replace my brain.
What doesn't work: AI-generated images for promo. I tried Midjourney and DALL-E for episode artwork and it all looked generic or weird. Hands are still a disaster. I'd rather use Canva with my own photos or pay a local designer $50. Also, I won't use AI voices (like ElevenLabs) for ads or intros. That crosses my ethical line — my voice is my brand. If I'm not saying it, it's not going out.
Here's my rule: AI is my intern, not my replacement. It does the first draft of boring stuff (show notes, email summaries, transcript cleanup) so I can spend time on what matters — the actual conversation, the music curation, the relationships. If you're using AI to avoid doing the creative work, you're doing it wrong. If you're using it to get past the tedious work faster, you're on it.
Cost breakdown: ChatGPT Plus is $20/month, Riverside transcription is included in my $19/month Creator plan. I'm not paying for Midjourney anymore ($10/month saved). Total AI spend: about $20/month, saves me 8-10 hours a month. That math works for me.
One more thing: always disclose when something is AI-assisted if it's public-facing content. I'll drop a line like 'show notes drafted with AI, edited by me' in my email footer. Transparency builds trust. Your audience isn't dumb — they'll know if you're trying to fake it.
If you're gonna try this, start small. Pick one repetitive task (transcription, show notes, email drafts) and test AI on that for a month. See if it actually saves you time or just creates more work. Then decide if it's worth the $20. Don't jump into every new tool just because it's hyped.
Drop a comment or DM me on IG @ninacapone — what's one task eating your time that you'd want an AI breakdown for? Let's figure it out together.