Developers launch SwitchOps AI integrating specialized agents with human review

On 2026-11-07, a developer introduced SwitchOps AI on DEV Community, a project built for the Flip the Switch Hackathon. The system is designed to address a familiar problem: when a customer asks for a proposal, the path from that request to a finished document pushes work through many hands. Someone has to pull up the customer's history, research the requirements, write a draft, review it, send it out, and remember to follow up. SwitchOps AI connects all of those steps in a single workspace instead of leaving them scattered across people, tools, and documents. Six specialist agents divide the labor between them, covering history retrieval, requirements research, drafting, review, sending, and follow-up. Reasoning is provided by Google Gemini, and the agents exchange messages through a Switch server. The architecture keeps humans firmly in the loop. Rather than letting the agents act fully on their own, the workflow is set up so that a person retains final control over the send action. The agents handle the drafting and the coordination around it, but nothing goes out to the customer without an explicit human decision. The project is presented as an early, exploratory build rather than a finished product, and the developer has linked the GitHub repository so other programmers can examine how the agents divide work, pass messages, and hand decisions back to a human. What makes the concept distinctive is the explicit separation between agents drafting the proposal and a person deciding whether it is actually sent, which frames it as a collaborative workflow tool rather than an autonomous pipeline.