Rohde & Schwarz AI Workplace

Why AI for test and measurement?

Today’s test setups involve more instruments, are more automated and run on tighter deadlines than they did even just a few years ago. Engineers manage complex, highly automated ecosystems under tight timelines.

A single validation cycle can span multiple instrument classes. In that complexity, one wrong setting does not just cause an isolated error; it can derail the whole schedule.

The tools built to address this complexity suffer from the same issues themselves:

• Too many commands: Instruments carry thousands of SCPI commands and native library methods, and the syntax shifts with every generation. This leads to too much time looking up parameters and too little time testing.
• Scattered know-how: Critical context sits across thick manuals, integrated development environment (IDE) scripts, live interfaces and raw traces, rather than being centralized in one location. Troubleshooting knowledge often exists only in the minds of a few senior engineers, who may not be available when a test fails.
• Slow log triage: When a test fails, engineers must manually cross-check signaling logs, radio-frequency (RF) captures and power traces across several layers. This is slow, error-prone process that can delay time to market.
• Moving standards: Standards across all industries are constantly evolving, and there are always new releases. Keeping test scripts, parameters and pass/fail rules current is tedious, ongoing work.

This is where AI-powered test and measurement can make a difference. It can fundamentally address this complexity and repetitive manual labor, not by answering more questions but by carrying out real tasks.

Meet the Rohde & Schwarz AI Workplace

The Rohde & Schwarz AI Workplace connects AI agents to your daily engineering work through two building blocks: (1) a Model Context Procol (MCP) server for instrument control and (2) AI agent plugins.

MCP works like a secure, standardized bridge. It lets an AI agent safely reach test and measurement instruments, SCPI interfaces, measurement apps, local files, application programming interfaces (API) and logs. The plugins add the domain expertise, turning a general AI model into a specialist that understands SCPI syntax and your specific instrument.

• Trusted context: AI workflows are connected to curated documentation, instrument manuals, SCPI references and customer-specific material, so agents surface the right information at the right moment. This eliminates the friction associated with manual look-up.
• Optimized runbooks: Predefined prompts, procedures and best-practice workflows keep agent behavior consistent and repeatable.
• Specialized skills: Plug-and-play capabilities for instrument control, SCPI sequence generation, driver interaction and log parsing.
• Guardrails and safety: Built-in operational limits, safety checks, security rules and approval gates keep automation predictable, auditable and safe for production environments.

How it works

The Rohde & Schwarz AI Workplace has four layers:

1. Language model: An LLM learns from huge amounts of text and code, which gives it a broad base of language patterns and general knowledge. With RAG, an LLM looks up something before it answers, instead of relying only on what it learned during training.

2. Verified technical context: The LLM pulls in relevant context (manuals, command references, specifications, etc.) and uses that to generate its response. This keeps answers tethered to official Rohde & Schwarz documentation and prevents hallucination.

3. MCP connections to instruments: MCP connects AI agents to your Rohde & Schwarz instruments and the wider engineering stack. Standard MCP servers let agents reach approved tools, control instruments, run SCPI commands, pull measurements and work with the IDE, without compromising security.

4. Agent harness: There is a control layer that turns reasoning into reliable action. It plans a step, runs it, checks the result and decide the next step. Session state, approval gates and guardrails help the agent stay on task and avoid repeating itself.

Available today on the CMX500

The Rohde & Schwarz AI Workplace is available today on the CMX500 radio communication tester. It is the first platform with the full assistant set: ScriptAssist, TechAssist and LogAssist.

What's next for AI-powered test and measurement?

The CMX500 is only the beginning. The same agentic pattern that works in the CMX500 today will extend across the entire Rohde & Schwarz portfolio.

More assistants for the CMX500

SpecAssist is planned for later this year. It will give agents grounded access to 3GPP specifications, including exact clause wording and ASN.1 definitions, so answers stay accurate instead of paraphrased.

More instruments (planned)

• Vector network analyzers: Automation of multiport calibration and de-embedding, checking responses along the way to catch setup errors early
• Signal generators and spectrum analyzers: Repeatable setups for modulation, multicarrier and phase-noise configurations, with waits and triggers to keep fast hops and synced carriers consistent
• Oscilloscopes: Automation of serial bus decoding, advanced triggering and deep-memory capture, so fleeting transients are not missed
• Power supplies: Safe, policy-checked power sequences and time-aligned current logging to avoid harmful sequences and surface bugs faster

AI extensions of existing Rohde & Schwarz tools (planned)

Beyond new assistants, Rohde & Schwarz is extending agentic workflows into tools you already use around the CMX500: CONTEST, CMsequencer, the installation manager and the report manager.

Flexible deployment for your environment

Every engineering environment is different, so the Rohde & Schwarz AI Workplace offers deployment options to match your security and integration needs.

• Enterprise and on-premises: On-premises MCP solutions for Mac or Unix, for teams that need deep integration and full data control
• Cloud trials: Time- or feature-limited cloud offerings for quick evaluation

Availability varies by assistant and instrument. Not every deployment model supports every AI Workplace tool or Rohde & Schwarz instrument.

Live instrument control, for example, works only with on-premises deployments and the standalone installer. Rohde & Schwarz confirms the right combination for your setup before you commit.