In yesterday’s campfire we talked (or at least, I did) at length about using LLMs and writing prompts. Here’s how I generate the show notes from the campfire zooms…
I grab a raw transcription from my <otter.ai> account. Then I pass that raw transcript to OpenAI’s Chat-GPT (v5 if you’re keeping score), with this prompt…
Mistakes matter a lot to me. Do not invent anything. Use only information from within the transcript provided. Work carefully and follow my instructions carefully. I prefer precision rather than speed. At all times, treat me like an expert. Only pause to ask me for clarification if a specific aspect of the instructions is ambiguous or contradictory.
Always check your work to ensure it follows all of my instructions before submitting it to me.
Perform the following 3 tasks.
Task 1: Show a heading with the word “Summary”. Leave a blank line. Write one paragraph summarizing the main topics of the conversation so that other topic experts can decide if they wish to read farther. Mention at least three topics which were discussed. Write the paragraphs in present tense. Do not bother including a summarizing sentence at the end of the paragraph.
Task 2: Show a heading with the word “Takeaways”. Leave a blank line. Write a list of takeaways from this conversation. Each takeaway must be on a separate line, with a blank line between each. Include as many unique themes or topics as you can. Do not number the list. Do not put bullets or any other markers at the start of each item in the list. Do not use any bold text. Each takeaway must begin with a phrase, then an em-dash surrounded by single spaces, and then a sentence (or sentence fragment) that describes what was covered in the conversation regarding that takeaway. Do not mention any of the conversation’s participants by name in the descriptions.
Task 3: Show a heading with the word “Resources”. Leave a blank line. Write a list of resources (people, contact information, places, books, videos, movies, web sites, etc.) mentioned in the conversation. Each resource must be on a separate line, with a blank line between each. Include as many resources as you can and provide a URL if possible. Do not number the list. Do not put bullets or any other markers at the start of each item. Do not use any bold. Each resource must begin with a phrase (hyperlinked if a URL is known), then an em-dash surrounded by single spaces, and then a sentence (or sentence fragment) describing the resource in detail and the context in which it came up in the discussion.
It spit out the following. I then spend a few minutes tweaking—in this case, I just found a few URLs for things and updated the links in the Resources section. A few things it didn’t do correctly: It didn’t actually link up the URLs in the resources, and it misspelled Forte’s first name…
Summary
The conversation covers the development of a potential new feature in a podcasting community, where members could be paired as peers or collaborators, exploring names such as “fellow,” “podmate,” and “mic mate.” It also discusses strategies for podcast management, including how to handle hiatuses, intros and outros, and filler episodes. A significant portion examines the use of large language models for writing assistance, prompt engineering, essay generation from podcast transcripts, and differences between “vibe coding” and more structured programming practices .
Takeaways
Community perk design — restricting attendance at campfire-style calls to paying members is positioned as a special benefit
Peer matching ideas — suggestions include pairing podcasters for ongoing collaboration without the heavy structure of masterminds
Naming challenges — explored terms like buddy system, fellows, podmate, mic mate, counterpart, and kindred voices to balance clarity, professionalism, and emotional resonance
Cultural references — analogies to Instagram’s photo sharing, WhatsApp’s messaging, and Ravelry’s knitting pattern sharing illustrate the concept of “killer features” in communities
Concerns about exclusion — noted that matching systems can feel like high school dances or picking playground teams, leaving some unpaired
Podcast hiatus management — debated whether podcasters owe listeners explanations for breaks and whether to use filler or “in-between” episodes
Episode intros and outros — contrasted polished introductions with minimalist starts and debated the long-term impact of embedded announcements
Use of AI tools — described workflows using ChatGPT and Claude for writing prompts, scripts, and essays, with specific challenges around memory limits and consistency
Prompt engineering process — explained using multiple LLMs in tandem, revision control systems, and exporting chat histories to refine and preserve prompts
Distinction in coding styles — differentiated “vibe coding” (under-specified, AI-driven) from structured programming with careful design
Exporting AI chats — shared technical methods for extracting full chat histories from web interfaces using DOM manipulation and JavaScript injection
Resources
LSU’s Tiger Stadium (Death Valley) — mentioned as a major venue for college football and the site of a game experience with 102,000+ fans
World War II Museum, New Orleans — described as “better than Smithsonian quality” and a highlight of a visit
Ravelry — https://www.ravelry.com — cited as a knitting community whose “killer feature” is sharing and modifying knitting patterns
Podmatch — https://podmatch.com — referenced as a platform that matches podcast hosts and guests, but with limited community engagement
Podcast Besties — described as a project for podcasters to cross-promote shows through feed drops and newsletter sharing
All Ears English podcast — https://www.allearsenglish.com — highlighted as a top-ranked English-learning podcast with millions of downloads
Claude.ai — https://claude.ai — used for searching and writing assistance, with some advantages over ChatGPT noted in the discussion
Thiago Forte — https://fortelabs.com — author referenced for his commentary on prompt engineering and differences between ChatGPT and Claude
Zencastr — https://zencastr.com — identified as the service providing raw transcripts for podcast episodes used in essay generation experiments