Particle

About Particle

Particle is a podcast intelligence API that transcribes, diarizes, and speaker-identifies podcast episodes shortly after they air, covering more than 100,000 shows with roughly 19,000 episodes processed daily. It provides full transcripts with speaker labels and word-level timestamps, entity extraction (people, companies, topics), semantic and keyword search across transcripts, guest and topic metadata, brand-safety and political-bias ratings, and rankings across the Apple Podcasts Top 200 in 130+ verticals. It also offers sponsorship intelligence that detects and maps ad reads to canonical brands, including placement type (pre-roll vs. mid-roll, host-read vs. pre-recorded) and an ad leaderboard.

The product is aimed at developers and agent builders: it exposes a REST API and an MCP server that connects to clients such as Claude Code, Claude Desktop, Cursor, and VS Code, and integrates with third-party platforms including Exa Connect and Lightfield. It supports configurable alerts (as-it-happens, daily, or weekly digests) delivered via email, Slack, or webhooks when tracked entities, keywords, or guests appear in new episodes, and provides embeddable HTML clips with synced audio and text for individual segments.

Key features

• Transcribes, diarizes, and speaker-identifies 100,000+ podcasts within minutes of airing
• Full-text search across transcripts by entity, keyword, or semantic match
• Sponsorship intelligence tracking ad reads, placements, and canonical brand mapping
• Real-time, daily, or weekly alerts for tracked entities, keywords, or guest appearances
• Rankings across the entire Apple Podcasts Top 200 in 130+ verticals
• MCP server for connecting to Claude, Cursor, and other AI agents

Frequently asked questions