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
"Everyone in the senior leadership team at Brex participates in continuous discovery monthly to build customer empathy."
Ned Dwyer Emily Stewart James WallisThe Intersection of Design and ResearchOps
September 24, 2024
"Most of our designers are contractors; they have limited voice because they’re bound by contracts."
Michael LandEstablishing Design Operations in Government
February 18, 2021
"How do we train AI models to prevent bias and preserve privacy, especially considering disability data is often seen as medical data?"
Sam ProulxTo Boldly Go: The New Frontiers of Accessibility
November 18, 2022
"We run our UX department as a business, with responsibilities, requirements, and expectations."
Vasileios XanthopoulosA Top-Down and Bottom-Up Approach to User-Centric Maturity at Scale
January 8, 2024
"Aligning content to audience expectations requires understanding their intent through market research."
Andrew Custage Michael MallettThe Digital Journey: Research on Consumer Frustration and Loyalty
March 29, 2023
"By limiting our classes to six or seven students, we could give better quality feedback."
Marjorie Stainback Kelsey KingmanTransforming Strategic Research Capacity through Democratization
October 24, 2019
"Connection is the energy that exists between people when they feel seen, heard, and valued."
Alla WeinbergCross-Functional Relationship Design
December 6, 2022
"Sometimes the loudest voices dominate, but asynchronous collaboration helps give quieter voices a chance to speak."
Jerome “Axle” BrownHow to Use Self-Directed Learning to Ensure Your Research Insights are Heard and Acted Upon
March 11, 2021
"The hiring process typically includes two manager screens and a final interview with project presentations."
Jen Crim Jess Quittner Saritha Kattekola Alex Karr Gurbani PahwaCulture, DIBS & Recruiting
June 11, 2021