A solo developer going by the handle fn2 posted to Hacker News on August 14, showcasing a Claude Code plugin that lets users query over 10 million earnings-call embeddings directly from their terminal. The project targets analysts and developers who want instant access to corporate financial sentiment without firing up a Bloomberg terminal or wrestling with traditional database queries.

How It Works

The system embeds earnings call transcripts into vector space, enabling semantic search across quarters of investor calls. Users type natural language questions like 'What did Apple say about AI infrastructure in Q3 2024?' and get back relevant passages with source citations. The plugin integrates directly with Anthropic's Claude Code environment, treating a decade of corporate financials as just another context window.

Background and Motivation

The developer says they've been running LLM-powered queries against a database of earnings transcripts since around 2025, initially using it for personal market research. Growing frustrated with the workflow, they built out a proper platform with additional financial tools and integrations rather than cobbling together scripts each time.

Technical Approach

Vector embeddings enable semantic similarity matching across the corpus of earnings calls from publicly traded companies. This approach outperforms keyword search for finding nuanced mentions of strategy shifts, management tone changes, or forward-looking statements that might not appear in traditional search indexes. The backend infrastructure handles embedding generation and retrieval while Claude Code provides the conversational interface.

Key Takeaways

  • 10.6 million embedded earnings-call segments searchable via natural language
  • Direct Claude Code integration eliminates context-switching for developers and analysts
  • Built by a solo developer as an internal tool that evolved into something shareable
  • Useful for market researchers, quant teams, and anyone tracking corporate narrative shifts

The Bottom Line

This is the kind of tooling that used to require enterprise budgets and data science teams. A single developer with some API access and vector database know-how just democratized earnings analysis for anyone running Claude in their workflowβ€”it's exactly how infrastructure should evolve.