We put together a comparison between OpenAI's AgentKit and FlowFuse:
The key difference: AgentKit is built for AI agents in the digital realm - chat, workflows, all within the OpenAI ecosystem. FlowFuse approaches it differently with MCP agents that run at the edge alongside your physical devices.
The edge piece is what makes it interesting for industrial use cases. When you're dealing with sensors, machines, and equipment, having agents make decisions right where the data is generated cuts out the cloud round-trip latency. Plus you get all the Node-RED capabilities for connecting to industrial protocols and managing fleets of devices.
Thoughts on this? Would love to hear if this resonates with what you're building or if we're missing key points in the comparison.