As an acquisitions editor at O’Reilly, I spend appreciable time monitoring our authors’ digital footprints. Their social media posts, talking engagements, and on-line thought management don’t simply replicate experience—they immediately impression e book gross sales and reveal promotional methods price replicating. Not surprisingly, a few of our best-selling authors are social media experts whose posting output is staggering. Maintaining with a number of superposters throughout platforms shortly turns into unsustainable.
I not too long ago constructed an AI answer to handle this problem. Utilizing Relay.app, I created a easy workflow to scrape LinkedIn posts from one writer (let’s name her Bridget), analyze them with ChatGPT, and ship me weekly e-mail summaries about her posts and which received probably the most consideration. The primary aim was to observe what she mentioned about her e book, adopted by thought management in her subject. The setup took 5 minutes and labored instantly. No extra periodically reviewing her profile or worrying about lacking vital posts.
However by the second abstract, some limitations turned obvious. Sorted by likes and impressions with generic summaries, each LinkedIn publish was receiving the identical therapy. I had solved the knowledge overload downside however now wanted a method to extract strategic perception.
To repair this, I labored with Claude to show the immediate into one thing nearer to an agent with fundamental decision-making authority. I gave it particular targets and choice standards geared toward shedding mild on promotional patterns that aren’t at all times simple to observe, not to mention analyze, in a flurry of posts: autonomously choose 10–15 precedence posts per week, prioritizing direct e book mentions; evaluate present efficiency in opposition to historic baselines; flag uncommon engagement patterns (each constructive and unfavorable); and robotically modify evaluation depth primarily based on how posts are performing.
The brand new report now offers deeper evaluation targeted totally on posts mentioning the e book, not simply any well-liked publish, together with strategic suggestions to enhance publish efficiency as an alternative of “this had probably the most likes.” Suggestions are sorted into short-term and long-term promotion concepts, and it has even proposed testing novel methods akin to posting quick video clips associated to e book chapters or incentive-driven posts.
The report isn’t good. The historic evaluation isn’t fairly proper but, and I’m engaged on producing visualizations. On the very least, it’s saving me time by automating the supply and evaluation of knowledge I’d in any other case need to get manually (and presumably overlook), and it’s starting to supply a place to begin for understanding what has labored in Bridget’s promotional program. Over time, with additional work, these insights may very well be shared with the writer to plan promotional campaigns for brand new books, or integrated into bigger comparisons of promotional methods between authors.
Whereas engaged on this, I’ve requested myself: Is that this an AI-enhanced automated workflow? An agent? An agentic workflow? Does it matter?
For my functions, I don’t suppose it does. Typically you want easy automation to seize data you would possibly miss. Typically you want extra goal-directed, versatile evaluation that leads to deeper perception and strategic suggestions. Extra of a useful assistant working behind the scenes week after week in your behalf. However getting caught up in definitions and labels is usually a distraction. As AI instruments grow to be extra accessible to everybody within the office, a extra priceless focus is present in constructing options that deal with your particular issues utilizing these new instruments—no matter you would possibly name them.
