@sfdata
San Francisco Data is building a marketplace for AI training data derived from production traces (real-world AI system usage). They propose capturing chronologies of AI working on actual tasks to create training datasets as models surpass human expert capabilities. The concept is technically coherent and addresses a real problem in frontier AI development, but the account shows minimal public activity (323 followers, no tweets visible, created March 2026) making it impossible to verify team, traction, or product status.
AI Analysisneutral
San Francisco Data is building a marketplace for AI training data derived from production traces (real-world AI system usage).
They propose capturing chronologies of AI working on actual tasks to create training datasets as models surpass human expert capabilities.
The concept is technically coherent and addresses a real problem in frontier AI development, but the account shows minimal public activity (323 followers, no tweets visible, created March 2026) making it impossible to verify team, traction, or product status.
Green flags: Addresses genuine technical problem in AI training data sourcing · Clear thesis on production traces as next-generation training data · Thoughtful long-form content on their approach
Red flags: Zero public tweets despite 7-month-old account raises questions about legitimacy · No visible team information or credentials · No evidence of working product, partnerships, or customers · Extremely low engagement (323 followers) with no community building · Could be vaporware or whitepaper-only project
Recent tweetsSee all on 𝕏 →
No recent originals to show — see the account on 𝕏.
Signal Timeline

Score breakdown0–100
Watching for additional signals.