Our co-founder Jacob Graubæk Houlberg recently joined Viraj Acharya on the VENTURES podcast to talk about how data can be used to identify early-stage startups before they become obvious.
The episode, titled The Niche of Early Stage Sourcing, covers how Evertrace connects fragmented signals across trade registries, GitHub, domains, patents, and more to map founders to what they are actually building.
Finding founders before they are visible
The core challenge in early-stage venture capital is that the best opportunities are often invisible when they matter most. Founders are building in stealth, incorporating companies quietly, pushing code to GitHub, registering domains, and filing patents, all before any public announcement or fundraise.
These signals exist, but they are scattered across dozens of unconnected sources.
During the conversation, Jacob explained how Evertrace brings these fragments together into a single feed. Rather than functioning as a static database, the platform is structured more like a real-time feed, surfacing what changed from yesterday to today. Investors check in each morning to see which new founders and companies have appeared overnight.
How AI powers the detection engine
A recurring theme in the episode was the role AI plays in making this possible at scale.
Evertrace has made over 200 million LLM calls, each answering small, specific questions: are these two names the same person, does this activity suggest a new company is forming, is this signal meaningful in context. The approach is to break massive data problems into small sub-problems that language models can solve reliably.
Jacob also discussed how improving AI models directly improve the product. As models get better at entity resolution, classification, and pattern matching, Evertrace's detection capabilities improve in tandem, without needing to rebuild the underlying infrastructure.
A growing wave of founders
The conversation also touched on a broader trend: the number of founders is increasing rapidly. AI-native development tools are lowering the barrier to building, big tech layoffs are pushing experienced operators to start their own companies, and experimentation has never been cheaper.
For investors, this means the sourcing problem is only getting harder. More founders means more signal, and more signal means greater need for infrastructure that can filter and prioritise what matters.
What we discussed
The full episode covers a wide range of topics, including:
- how Evertrace detects new founders across fragmented data sources like trade registries, GitHub, patents, domains, and co-founder searches
- why the platform is structured as a real-time feed rather than a traditional database
- the role of AI in entity resolution and signal classification at scale
- how the growing number of founders is making systematic sourcing more critical
- the decision to bootstrap Evertrace and how the company has grown to over 15% month-over-month revenue growth
- recent acquisitions including Morphais, Seedpoint, and Whisper AI
- how MCP integration allows AI agents to connect directly to Evertrace
Watch the episode
If you are interested in how investors can use data to find founders before anyone else, this episode is well worth a watch.
A big thank you to Viraj Acharya for the great conversation and for having us on the show.
You can watch the full VENTURES episode below.
The founder detection engine for VCs
AI-powered founder detection with unmatched coverage.
Book a demo