Agent for book brainstorming
I think I was subconsciously inspired by Federico Viticci’s post when I tried this prompt with ChatGPT Agent last night:
I created book.micro.blog and www.manton.org. I want to look through longer blog posts (at least a couple paragraphs) on my blog that are about social media and blogging. Find about 50 of the most interesting posts and then look at book.micro.blog to see if those topics were very similarly written about in a chapter. If they were not covered, keep track of that and at the end make a short report of each blog post and why it would be good to write about for an updated version of the book, and where it might fit.
After churning on it for 31 minutes, the result was this report. It’s very good. Here are just a few examples:
6 Jul 2023 – “Adam Mosseri on why ActivityPub” – Quotes Instagram head Adam Mosseri explaining the value of open protocols and decentralisation. Including this in ActivityPub would illustrate mainstream interest in federated social networks.
25 Apr 2025 – “Bluesky downtime” – Explains why Bluesky’s downtime sparked questions about decentralisation and compares Bluesky and Mastodon infrastructures. Adding this to Decentralization would help readers understand federated resilience and the trade‑offs of different protocols.
15 Feb 2024 – “Mastodon and public data” – Comments on debates around bridging ActivityPub to Bluesky and warns that insular decisions might isolate Mastodon from the open web. This update would enrich Mastodon or WebSub by explaining why openness benefits all platforms.
I won’t use everything that the agent found, but there are a bunch that are worth at least checking. I knew that I had blogged about things that I should’ve edited into the book too, while I was thinking about the topic. It would’ve been extremely time-consuming to figure out how to create a list like this manually. Probably a full day or two of work.
These AI tools are getting scary good very quickly.