Hello everyone,
After attending a talk at IEEE Quantum Week about El Agente, an autonomous AI system and agent hierarchy for computational chemistry ( Acceleration Consortium 2025 ), I started thinking about how AI agents could fit into the Juice platform.
One possible approach would be to have agents specialized around different roles:
Lab Manager
- Monitors the health of instruments, services, and infrastructure.
- Tracks system status, detects anomalies, and helps prioritize tasks.
- Naturally connects to the System Overview and Task Manager.
Experimentalist
- Assists with experiment design, execution, and analysis.
- Recommends next steps based on results and experiment data.
- Could leverage AI-driven workflows such as Ising model (AI Models & Framework for Quantum Computing | NVIDIA Developer).
Juice Developer
- Understands the Juice codebase, APIs, and documentation.
- Helps users develop extensions, troubleshoot integrations, and navigate the architecture.
- Acts as an interactive technical copilot for the platform.
The more interesting question to me is not whether to add AI agents, but how users should interact with them.
Should agents be:
- Dedicated personas with clear responsibilities?
- A single assistant that can switch contexts?
- What tools and skills are most useful?
- Embedded directly into workflows and dashboards?
- Autonomous agents that can execute actions, or mostly advisors that recommend actions?
I’m curious to hear what the community thinks. If AI agents became a native part of Juice, what would be the most valuable use case for you, and what would the ideal user experience look like?
Looking forward to the discussion!
