Making spoken conversations searchable for AI systems

An artificial intelligence startup founded by former engineers from Twitter is shifting its core focus toward indexing spoken conversations. The company introduced a platform designed to transcribe podcast audio and interpret its meaning, allowing users and automated systems to extract key quotes and highlights from audio files.

Providing context for automated agents

Most software tools crawl the web and focus exclusively on text, leaving autonomous agents largely blind to audio content. The new platform bridges this gap by transcribing and indexing over 130,000 podcasts across 135 verticals, including all of the top 200 shows on Apple. The system processes and adds roughly 20,000 new episodes to its index every day.

The underlying technology does more than convert speech to text. It includes speaker labels, tracks mentions of specific people and companies, and understands the broader topics discussed. Users can configure automated alerts delivered via email, Slack, or webhook to notify them when specific guests appear or when particular topics come up.

Applications and pricing

Beyond basic search and transcription, the platform provides dedicated tools for tracking podcast advertisements, listener ratings, sponsorship data, and political bias. While the service includes a web interface, the primary offering is a developer application programming interface designed to let external businesses and software agents query podcast intelligence programmatically.

Early adoption is coming from financial institutions and hedge funds seeking alternative data sources for their internal systems. Consumer tiers are priced at $29 per month for a single seat, while business plans cost $399 per month and include 20 seats. Custom pricing is available for heavy users of the interface, and the team plans to expand coverage to other audio formats like video clips in the future.