The LavaCon Content Strategy Conference | 27–30 October 2024 | Portland, OR
Jennifer Swallow

Jennifer Swallow is a controlled language expert who has brought her passion for simplified technical English and a good template to many different roles. Her twenty-year career spans a variety of industries, including standardized testing, medical devices, sailing, and cybersecurity. At Splunk, she runs a customer success website that documents use cases, best practices, and tips from field personnel. She’s highly skeptical of AI applications, mainly because of poor training data, but she’s also interested in using it more. When not at work, Jennifer can be found scaling rugged mountains, running ultramarathons, or walking dogs at the Albuquerque City Shelter.

The Beauty and the Beast: A Relationship with ChatGPT

Co-presented with: Shane Newman

For many content creators, ChatGPT is a Beast, a looming monster threatening to take away the livelihoods of the Beauties, the highly-skilled knowledge workers. Yet we can’t avoid it. We are being forced to confront it head-on. Rather than allow it to imprison us, what if we lean into what it has to offer? What prince might we find hidden beneath the scary surface? This talk is a case study on how a crowdsourced knowledge base, Splunk Lantern, is using articles generated through a custom AI model to provide more self-service content to customers. Using ChatGPT for first drafts allows both the contributors and the content management team to build out content faster to support more Splunk programs and answer more customer questions. But the process hasn’t been without its challenges. Just as Belle and the Beast had a few false starts while they tried to understand each other, so too has working with ChatGPT been an iterative process. But the result is a benefit to Splunk Lantern users and Splunk customers as a whole.

In this session, attendees will learn:

1. How to create an AI model that uses the company’s public-facing docs to generate new content based on a specific outline

2. What human editors caught while editing

  • Grammar and style issues
  • Wordiness and nonsense phrases
  • Content organization challenges

3. How the process to retrain the model based on editorial feedback worked (and didn’t work)

4. What benefit this AI/human partnership is bringing to the company

  • Enhanced content quality and efficiency
  • Improved knowledge management strategy
  • Better support and user experience through targeted self-help articles