Networking with agents: Put them in the right conversations with Tailscale

Happily handing over all your private data to a shady tiger in a trenchcoat - seems like an apt visual for the current state of AI.

I worked for a company once where we ran this dodgy shipping software on prem that integrated with our backend via SQL access. When there was an issue, their techs would rdp to a server and run this little VB app that turned out to be a dialog box that could run arbitrary SQL code against our production database.

I trust that more than this nonsense. WTF are we doing?

So is inference happening in network or not?