a "multiplayer" AI workspace where you watch agents do your actual work
, and I've been building something I'm calling FlowRoom. Roast it.
The idea: most AI tools are one person typing to one chatbot. FlowRoom is a shared room where multiple people and multiple AI agents work together on real, long-running operations and anyone can drop into the same live session to watch an agent work, redirect it mid-task, or hand it off to someone else.
First use case I'm targeting is industrial shift operations (think mining/processing plants big here). A fault gets reported, an agent picks it up, extracts the details, assigns a technician, flags anything expensive for human approval, and rolls it into the shift report. Humans stay in the loop for anything irreversible.
Long-term vision is that it's not just mining ,it's a customizable platform where any team (NGOs, sales, legal, schools) builds their own workflow without code.
What I actually want roasted:
- Is "multiplayer AI agents" a real need or a solution looking for a problem?
- Is starting with mining/industrial ops smart, or am I picking a brutal first customer?
- Everyone and their dog is building "AI agent platforms" right now. Why would this survive?
Be honest, I'd rather get torn apart here than by the market.
1 upvote