# Unsupervised Topic Discovery with Text Embeddings

DevFeed: [Unsupervised Topic Discovery with Text Embeddings](<https://devfeed.tech/articles/unsupervised-topic-discovery-with-text-embeddings-65153.md>)

Original publisher: [Read original article](<https://eng.wealthfront.com/2026/10/05/unsupervised-topic-discovery-with-text-embeddings/>)

Author: Eli Berg

Published: 2026-10-05T17:46:10Z

Content type: tutorial

Language: en

Sources: [Wealthfront](<https://devfeed.tech/sources/wealthfront.md>)

Topics: [Embeddings](<https://devfeed.tech/topics/embeddings.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>)

Tags: [ai-research](<https://devfeed.tech/tags/ai-research.md>), [algorithm](<https://devfeed.tech/tags/algorithm.md>), [automation](<https://devfeed.tech/tags/automation.md>), [business](<https://devfeed.tech/tags/business.md>), [data](<https://devfeed.tech/tags/data.md>), [embeddings](<https://devfeed.tech/tags/embeddings.md>), [features](<https://devfeed.tech/tags/features.md>), [text-embeddings](<https://devfeed.tech/tags/text-embeddings.md>), [wealthfront-engineering](<https://devfeed.tech/tags/wealthfront-engineering.md>)

## AI overview

Wealthfront describes a pipeline for discovering topics in support-call notes: text embeddings are reduced with UMAP, clustered with HDBSCAN, and labeled using an LLM. It then discusses hierarchical clustering, using labeled examples to classify new notes, and retraining to detect emerging topics while managing label drift.

## Source excerpt

At Wealthfront, a core pillar of our business model is addressing our clients' needs proactively, efficiently, and with minimal unnecessary toil. As our client base grows, and as we ship an ever broader suite of products and features, the breadth of issues our support and operations staff have to address grows as well. Our preference... Read more