Potscan

Arvid built Potscan to solve his own problem: understanding what people are saying about brands and ideas across the massive landscape of podcast content. With 4 million podcasts broadcasting daily, the opportunity to turn "unstructured podcast chatter into competitive intelligence" was obvious. He wanted to create a platform where founders, PR experts, and marketing teams could monitor mentions in real time and act on customer insights, PR opportunities, and competitive threats.

The platform started with fundamental infrastructure challenges. Potscan needed to index and search across "dozens of millions, almost 50 million podcast episodes" with transcripts. Arvid initially built the search system using Laravel and Meilisearch, a lightweight full-text search engine. The choice reflected his past frustration with Elasticsearch—a decade earlier (around 2015-2016), he'd been burned by its complexity. But as the dataset exploded in size, Meilisearch began to struggle with data ingestion despite its sub-millisecond query speed. About six months into the narrative timeline, Arvid made a critical decision: migrate to an OpenSearch cluster on AWS, a choice he would have rejected two years prior without AI assistance.

The most successful channel proved to be programmatic SEO. For 18 months, Arvid wasn't sure the effort was working, but patience paid off. Each new podcast page on Potscan.fm gradually built domain authority through backlinks from major publications like the Wall Street Journal and Forbes. The insight: podcasters themselves began linking to their Potscan pages in show notes and claiming their shows on the platform, creating a virtuous cycle of organic traffic and legitimacy. "The name has become almost synonymous with podcast analytics in certain circles," Arvid noted. The platform now receives hundreds of new users daily through search alone.

A critical product decision was adding an embeddable player, inspired by Transistor.fm. This feature drove additional backlinks and made Potscan essential infrastructure for podcast promotion. The integration with OP3, an open standard for transparent podcast analytics, proved equally valuable—it synchronized real download data into Potscan's database, improving the accuracy of estimation models across the entire platform.

On the automation front, Arvid built AI-drafted outreach emails that analyze each user's behavior and suggest the highest-impact next step. This semi-automated approach (10% manual, 80% AI, 10% manual refinement) helped him scale engagement without burning out. AI agents also proved essential for tackling Elasticsearch's DSL (domain-specific language)—a task he'd previously avoided entirely. "Agent decoding has helped me unlock capabilities that I would have never attempted on my own," he said.

Potscan has evolved into a multi-product offering. The core analytics platform monitors 4 million podcasts in real time. A secondary product, ideas.potscan.fm, uses AI to identify startup opportunities from expert discussions across hundreds of hours of podcast content. The data quality has become a competitive moat: better transcripts and integrations lead to better search results, which drive more backlinks, which improve domain authority in a reinforcing cycle. Arvid's shift toward semi-automated systems and AI-assisted development has freed him from day-to-day ops work while maintaining quality and personalization at scale. Patience—his initial 18-month SEO investment—proved to be a sustainable competitive advantage in a crowded analytics market.