Last August, I updated you on InPublishing’s first tentative steps into the world of AI and I thought it was time for another quick update.
In a nutshell, AI is beginning to bring tangible benefits for us.
As mentioned previously, we are using a simple prompt to convert Word documents to html to speed up the process of uploading articles to our website. By using this prompt (through the Claude AI platform), we save the time we would have spent manually formatting the text, thereby making our workflow more efficient. Having said that, the AI is not 100% consistent, so every output has to be eyeballed.
Our efforts to create an “Ask InPublishing” query tool have stalled. We were finding that the AI wasn’t always selecting the most relevant content from our archive when building its response and we struggled a bit with misattributed quotes. We think this is worth exploring further but for the time being, this bit of new product development is on the back burner.
And to those two applications, we’ve now added a third, marketing-related one, which has proved simple but effective.
Our key prospects are publishers.
When looking to build prospect lists, our own site is one of the best places to find them because we’re constantly writing about publishers in the articles and news stories we publish.
Previously, time permitting (which it often wasn’t) we would manually trawl through articles pulling out names.
Super clunky, time consuming, sporadic and inefficient.
We have now developed a very simple AI that does the work for us — it goes through each article and pulls out names and drops them into a database containing the following fields: (name, job title, company, leaver (Y/N) along with the ‘date published’ and the url of the original story, so we can randomly check the output.
We’ve embedded the process into our website and it has pulled 9,700 names (when deduped, that went down to about 5,500) from the last two years’ worth of articles. Development time aside, the list costed approximately $25 to compile.
The list will still need the celebrated human-in-the-loop (Rishi Sunak and Mel B were two of the names pulled) but the process of name extraction has been automated. The word ‘game-changer’ is over-used, but this feels like a significant step forward.
(On the subject of AI, in our recent AI Special Q&A webinar, we didn’t have time to answer all the attendee questions. We have now put that right and our panelists have given their responses to those questions, which you can read here.)
You can catch James Evelegh’s regular column in the InPubWeekly newsletter, which you can register to receive here.
