Any plans on making it possible to launch agents (say, via Slack), that run directly on self hosted machines? It would be useful for workloads that require specific environments (e.g. GPU environments hosted on Cloud Providers, that include the necessary system dependencies and hardware).
Example:
Launch an agent via Slack. It then runs on a Cloud VM on Google Compute Engine with Nvidia GPUs, then the agent can develop and run AI workloads (run image models, to evals, etc).
This would be very nice! Also would make sense security wise if you want agents to have access to security relevant data that are not allowed to leave your companies server.