1) AI got really good at some things like facial recognition and identifying people (see something like Lenso), where’s the moat here?
2) I get that first advantage has a moat in collected data, but can’t AI just get this from the source? Companies like Checkr and Truv use APIs to let candidates log into their own payroll accounts (e.g., ADP, Workday, Paychex, or Gusto), or the AI agent can just login to the government’s public court record websites for the data.
I don’t really see you address the existential AI risk there despite the very attractive numbers. Maybe that implementation takes too long to matter for FA’s valuation. Time will tell.
The court records exist at the county level and not enough are digital yet for Checkr to create a robust enough database to compete with FA. This will take at least a decade to become a threat. The write up contains a section that touches on this. No AI agent can log into public court record websites, many have no online portal and require a network of physical runners.
Yeah but realistically how many courts are like that? Major cities that probably account for 90%+ of population all have digital court documents
I think I mostly agree on the implementation time taking awhile, but not ten years, my best guess would be 3-5. Just a guess. But you already have AI interacting with websites, APIs, logins, etc., albeit governments move at a snail pace
Interesting piece. A few questions.
1) AI got really good at some things like facial recognition and identifying people (see something like Lenso), where’s the moat here?
2) I get that first advantage has a moat in collected data, but can’t AI just get this from the source? Companies like Checkr and Truv use APIs to let candidates log into their own payroll accounts (e.g., ADP, Workday, Paychex, or Gusto), or the AI agent can just login to the government’s public court record websites for the data.
I don’t really see you address the existential AI risk there despite the very attractive numbers. Maybe that implementation takes too long to matter for FA’s valuation. Time will tell.
The court records exist at the county level and not enough are digital yet for Checkr to create a robust enough database to compete with FA. This will take at least a decade to become a threat. The write up contains a section that touches on this. No AI agent can log into public court record websites, many have no online portal and require a network of physical runners.
Yeah but realistically how many courts are like that? Major cities that probably account for 90%+ of population all have digital court documents
I think I mostly agree on the implementation time taking awhile, but not ten years, my best guess would be 3-5. Just a guess. But you already have AI interacting with websites, APIs, logins, etc., albeit governments move at a snail pace