LAP: An Agent-to-Instrument Protocol for Autonomous Science

2026-06-02 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems · Depth: Expert, quick

Summary

The Lab Agent Protocol (LAP) is introduced as a new standard designed to connect AI reasoning agents with physical scientific instruments in autonomous science. This protocol addresses a critical gap, as existing agent interoperability standards like Anthropic's Model Context Protocol (MCP) and Google's Agent2Agent (A2A) do not model the stateful, safety-critical agent-to-instrument edge. LAP builds on A2A's peer-to-peer, discovery-first, task-lifecycle structure, adding four key physical-world primitives: an InstrumentCard for capabilities and limits, first-class reservation for exclusive locking, a safety-fence handshake with operator-confirmation tokens for hazardous operations, and a MeasurementResult schema ensuring physically typed (QUDT/UCUM), calibration-anchored, and reproducible results. The protocol specifies roles, a six-layer architecture, JSON-RPC methods, and state machines, while encapsulating existing device standards such as SiLA 2 and OPC-UA. It was published on 2026-06-02.

Key takeaway

For AI Engineers developing autonomous scientific laboratories, LAP offers a crucial framework to standardize agent-to-instrument communication. This protocol provides essential primitives for safety, state management, and data reproducibility, eliminating the need to custom-build interfaces for each instrument. Adopting LAP can streamline integration efforts, enhance operational reliability, and accelerate the deployment of robust self-driving research systems, ensuring safer and more reproducible experimental campaigns.

Key insights

LAP provides a standardized protocol for AI agents to safely and reliably control physical scientific instruments, filling a critical gap in autonomous science infrastructure.

Principles

• Autonomous science requires dedicated agent-instrument protocols.
• Physical-world operations demand explicit safety and reproducibility.
• Existing device standards can be encapsulated, not replaced.

Method

LAP defines a six-layer architecture, JSON-RPC method set, and task/safety state machines. It uses InstrumentCards for capabilities, reservations for locking, safety-fence handshakes for hazardous operations, and a MeasurementResult schema for data.

In practice

• Implement InstrumentCards for device capability description.
• Integrate reservation systems for exclusive instrument access.
• Utilize safety-fence handshakes for critical operations.

Topics

• Autonomous Science
• AI Agents
• Instrument Control
• Protocol Design
• Laboratory Automation
• Scientific Reproducibility

Best for: AI Scientist, AI Engineer, Research Scientist

Related on AIssential

• From Data to Theory: Autonomous Large Language Model Agents for Materials Science — Autonomous LLM agents can perform end-to-end scientific theory development, from equation selection to code execution and validation.
• Claw AI Lab: An Autonomous Multi-Agent Research Team — Claw AI Lab transforms autonomous research into an interactive, steerable, and reliable multi-agent laboratory environment.
• Quoting Karel D'Oosterlinck — AI can automate research due diligence and experiment setup, saving significant manual effort.
• Science Earth: Towards A Planet-Scale Operating System for AI-Native Scientific Discovery — A planet-scale runtime enables AI capabilities to self-organize for distributed, self-correcting scientific discovery.
• The Agent Protocol Handbook — Standardized protocols like MCP and AP2 are crucial for enabling autonomous AI agents to interact and transact securely.
• NIMO Controller: a self-driving laboratory orchestrator based on the Model Context Protocol — NIMO Controller unifies human and AI interaction with self-driving labs via the Model Context Protocol.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.