01
Podcast & YouTube Ingestion
IngestionContinuous intake of podcast and YouTube audio so new episodes are available to search shortly after they publish.
You've Been Mentioned
You've Been Mentioned is an audio intelligence platform built for PR teams, brands, and investors to track spoken mentions across the podcast and YouTube ecosystem.
We built the customer product and the production systems behind it: continuous ingestion of podcast and YouTube audio, transcription with WhisperX, and search across spoken mentions—so teams can find coverage and get alerted as it lands.
The Challenge
Overcoming critical bottlenecks in real-time execution.
PR leaders, corporate communications teams, and venture firms need to know the moment their brand, portfolio company, or competitor is named in spoken media. Traditional media monitoring was fundamentally built for the printed web: crawling articles and social media within minutes, while audio podcasts and YouTube discussions remained an unindexed black box for days.
Core friction points
- Unprecedented media volume: Tens of thousands of audio hours published every single day across fragmented RSS feeds, YouTube channels, and streaming platforms.
- Acoustic variability: Extreme variations in microphone quality, ambient background noise, overlapping speakers, and varied accents that break standard off-the-shelf transcription APIs.
- Critical latency constraints: A public relations crisis requires immediate response within minutes of an episode going live, making slow overnight batch processing unacceptable.
- Search precision vs. recall: Distinguishing between casual common words and exact enterprise brand entities (e.g. distinguishing 'apple' as a fruit from 'Apple Inc.').
Solution & Architecture
Engineered for speed, durability, and fault tolerance.
We built the customer product and the production audio pipeline behind it: ingest spoken media, transcribe it, and make it searchable at scale. Infrastructure is defined end-to-end in AWS CDK.
02
Speech-to-Text at Scale
TranscriptionProduction transcription with WhisperX, turning spoken audio into searchable, time-aligned text.
03
Transcript Search
SearchSpoken content indexed in OpenSearch so teams can find exact mentions and related discussion across the catalog.
04
Alerts & Tracking Product
ProductA full customer product for transcript search, mention alerts, and competitor tracking, running on AWS.
Core technologies:Next.js · TypeScript · Python · WhisperX · PostgreSQL · OpenSearch · AWS · AWS CDK
Outcomes & Impact
Measurable results in production.
The platform launched smoothly to enterprise launch partners and rapidly scaled to full production, continuously indexing millions of audio words daily with zero dropped feeds.
150k+Shows & channels indexed
99%Transcription accuracy
24/7Continuous audio ingestion
< 90sMedian publish-to-alert latency
- Over 150,000 active podcasts and YouTube channels monitored 24 hours a day, 7 days a week.
- Median time from podcast episode publication to customer inbox notification dropped to under 90 seconds.
- 99.9% uptime on continuous ingestion clusters processing terabytes of media monthly.
- Reduced client manual monitoring overhead by over 80% while capturing 4x more spoken mentions.
Start a project
Tell us about what you're building.
Whether you have a detailed technical specification or an early concept on a napkin, we're happy to review your requirements, talk through architecture, and see if we're the right fit.