- We used NotebookLM to turn years of support docs into a searchable knowledge base
- The support team can now instantly query everything, with links to the sources
- NotebookLM also made a podcast about our support history — hilarious, but not super useful!
This is flipping epic and probably one of the most best uses of AI we've found for Alphero work yet. We support a load of client apps / products many with literally years of history, context and background knowledge. As you'd expect, newbies on some codebases can need time to get up to speed. We thought we'd take a crack at using an AI tool (under the steely eyes of Brendon Price our principal security consultant).
We used NotebookLM and after exporting support history including documentation, design, support tickets and general commentary, requirements, app store feedback and other stuff into a new Notebook LM 'project' I can confirm #nailedit. Not only do we have an instant and actually remarkably useful knowledge base that the support crew can query on any topic related just to that project / product. It also links you back to the sources. All in a walled garden. This bit is great and we're going to do more of it.
Not nearly so useful but infinitely more entertaining was generating an AI podcast off the content ... the NotebookLM artificial hosts bravely approached telling the story of 7-10 years of support tickets and documentation for one product with artificial confidence, poise and an open mind. Albeit calling out the "wall of jargon" that must be navigated to go from lines of code to an app.
Personal favourite AI podcast quotes:
- "What about API specs! That one sounds particularly intimidating" 😱
- "the documents give us some clues about the process" 🤔
- "those unsung heroes [developers!] deserve a round of applause." 🙌
And the best of the bunch (in relation to a specific error report):
"the app is having an existential crisis or something". 😭
Anyhoo, we'll definitely be doing this again for other longer term projects. Maybe not the podcast bit.
