Launched this week
Wazear
A visual AI orchestrator where agents check their own work
4 followers
A visual AI orchestrator where agents check their own work
4 followers
Wazear is a local desktop AI agent orchestrator that runs structured pipelines including review agents that rigorously check the output of other agents. Whether the output is an implementation plan, architecture or actual code. You can interrupt at any moment and make adjustments as you see fit.
Hey, the YT video is not working.
@richard_andrews4 Thank you very much for the heads up. Sorry about that, fixed it now.
@vallar Hi. Visual AI orchestrator where agents check their own work - do they really self-check?
@julia_zakharova2
Hello and thank you for your question.
Yes, though this is dependent on what your pipeline looks like. If you use the sample agents that ship with the app and leave everything to default, yes. Wazear will spawn an agent and run it as say the planner. It produces its own output (the plan) based on a brief. The next agent running is a planner reviewer, it checks the work that the planner did. If something is wrong, the pipeline is looped back to the planner, the planner then checks the feedback, checks what it did and then corrects it according to the feedback.
Of course, if your pipeline is completely linear and no review loops are added, then agents don't ever check anything and it is just one agent handing to the next without any oversight.