The SaaS Podcast

Nate Sanders, Artifact

How This AI SaaS Solved the Cold Start Data Problem

While working at Pluralsight, Nate Sanders spent 70-85% of his time doing something most product leaders dread - manually synthesizing customer research data across departments. Teams were running thousands of customer interviews a year, but the synthesis was happening in glass conference rooms covered in sticky notes, with 10 people earning $150K+ spending two days on affinity mapping exercises. When the first large language model, BERT, emerged in 2018, Nate ran internal experiments and saw promising results. He and co-founder Trey spent seven months of nights and weekends building prototypes in React, testing half a dozen different product ideas before raising a small angel round from friends and family in the Salt Lake City area. But building an AI SaaS product created a painful chicken-and-egg problem. Artifact needed training data to fine-tune its models, but getting training data required having customers, and customers required a working product. Nate solved this by recruiting nearly a dozen design partners through his network, offering free product access in exchange for their data, feedback, and 30 minutes of weekly time. Each partner signed letters of intent with paid deposits of $1,000 to $1,500 - small enough to vanish on a corporate card, but enough to prove real commitment. With three to four active design partners collaborating weekly, Artifact closed its first $100K in ARR and raised a $5M seed round led by Josh Buckley of Buckley Ventures. But scaling this AI SaaS beyond those initial partners proved harder than expected. The team tried bottoms-up self-service, community building, dinner events in San Francisco, and conferences - spending $30K-$40K on events alone before realizing none of these channels attracted the right buyers. The breakthrough came when Nate discovered that outbound SDR-driven outreach to enterprise accounts produced dramatically better retention, higher ACVs, and stronger engagement than any other channel. Going down-market was tempting because smaller companies moved faster, but enterprise customers had 90% higher ACVs and significantly lower churn. Nate's AI SaaS playbook became clear: double down on a structured SDR-to-AE pipeline with commitment-based sales stages, replacing subjective pipeline assessments with concrete customer actions like sending data source lists and authenticating integrations.