Summary
Keeping large content repositories organized is an ongoing challenge. There's always new stuff coming in, and taxonomies evolve over time. Resource-strapped teams seldom have opportunities to re-categorize older content. It's a task well-suited for generative AI. Large language models have powerful capabilities that can help teams keep content organized at scale. Using LLMs in this capacity can lead to better user experiences and free team members to focus on more valuable efforts. This presentation explores two approaches for using LLMs to organize content at scale: 1) re-categorizing content using existing categories and 2) developing new categories from existing content. Both will be shown as proofs of concept alongside feasible next steps.
Key Insights
-
•
Maintaining organized large content repositories is often deprioritized despite evolving product portfolios, leading to reduced findability.
-
•
Search alone is insufficient for large repositories because users often lack context about what content exists.
-
•
Evolving taxonomies require periodic retagging of older content to keep it relevant and findable.
-
•
AI, specifically large language models like GPT-4, can automate taxonomy maintenance tasks significantly faster than humans.
-
•
Human oversight is critical to review AI-suggested tags to prevent hallucinations and ensure quality.
-
•
Developing new taxonomies requires analyzing content holistically, which exceeds typical LLM context windows.
-
•
Embedding databases help chunk and relate content statistically for clustering similar topics at scale.
-
•
Graph Retrieval Augmented Generation (Graph RAG) integrates knowledge graphs with LLMs to increase precision by leveraging semantic relationships.
-
•
RAG enables querying domain-specific content unseen by generic LLM training data, supporting bespoke AI assistants.
-
•
AI tools enable higher speed and new methods for content work but require new workflows and ongoing experimentation.
Notable Quotes
"Keeping large content repositories organized is an ongoing challenge that often gets deprioritized."
"Search alone doesn’t cut it, particularly when people don’t know what they’re looking for or lack context."
"If organizations don’t keep content organized, it becomes less usable and less useful over time."
"Large language models can help organize content faster and at much larger scale than people can."
"I used GPT-4 to retag 1,200 blog posts, reducing an estimated 10–12 hours of tedious manual work to about a third of that time."
"I made sure the LLM would only use terms from my predefined taxonomy, but it still introduced some of its own tags."
"Working with these tools requires new workflows and checkpoints to validate and tweak the AI's output."
"Graph RAG uses knowledge graphs instead of just plain text snippets, which really improves precision."
"I see these AI tools as augmenting human work, letting us focus on what matters most, not replacing humans."
"The best way to understand these technologies is to learn them hands-on, especially if your work involves language or taxonomies."
Or choose a question:
More Videos
"Don’t get stuck in the refinement stage too early."
Billy CarlsonIdeation tips for Product Managers
December 6, 2022
"If you haven’t done accessibility work yet, it makes sense to get started on mobile because you already have the tools in your pocket."
Sam ProulxMobile Accessibility: Why Moving Accessibility Beyond the Desktop is Critical in a Mobile-first World
November 17, 2022
"Enterprise problems are often engineering problems as much as design problems."
Dan WillisEnterprise Storytelling Sessions
May 13, 2015
"We are outnumbered by engineers maybe a dozen to one, but things are improving all around."
Michael PolivkaScaling Design through Relationship Maps
November 7, 2017
"The top LLM we tested got only 47% accuracy categorizing Reddit posts, which is sobering."
Karen McGrane Jeff EatonAI for Information Architects: Are the robots coming for our jobs?
November 21, 2024
"If you make rules, put numbers in them, and say them confidently, sometimes people actually do them."
Leisa ReicheltOpening Keynote: Operating in Context
November 7, 2018
"If you find someone not following values, start with a one-on-one to understand their point of view and realign together."
Kim Holt Emma Wylds Pearl Koppenhaver Maisee XiongA Salesforce Panel Discussion on Values-Driven DesignOps
September 8, 2022
"Change requires more than a single vision or a few projects; it takes careful orchestration throughout the organization."
Jacqui Frey Alison RandSetting the Table for Dynamic Change
October 24, 2019
"We have to learn to speak in the language of the business to get their attention and investment."
How to Identify and Increase your "Experience Quotient"
June 15, 2018
