The workspace is the place where an engineering team keeps its work in Plotune. You create a workspace, add projects to it, invite people with roles, and switch on the tools each project needs. RMS (requirements) and Stream (event history) are available today. Nexus and Bridge are hardware products of their own for now; adding them to a workspace the same way is coming.
Questions about Plotune
How the workspace and its tools fit together, and how teams use Plotune Nexus to run bounded, AI-ready workflows on real test hardware while keeping their data local.

About Plotune
RMS is the requirements tool in a Plotune workspace. It keeps requirements with revisions, review and approval, traceability links, test cases with recorded evidence, and baselines, and it exchanges requirement sets as ReqIF. AI agents can work in it through MCP, with roles and an audit trail. RMS makes no claim of compliance with any standard.
Open the project in your workspace and switch on the tool it needs, for example RMS or Stream. The tool then appears inside that project, and only the people who have access to the project see it. Switching a tool off again removes it from the project view. Nexus and Bridge are not added this way yet.
A local-first acquisition appliance captures and stores data next to the hardware by default, rather than streaming it into a vendor cloud first. Plotune Nexus is a managed local-first appliance on Plotune-defined hardware: captures and artifacts stay on the device, and the team chooses where outputs go, whether local files, SFTP, Google Drive, or OneDrive and SharePoint.
AI-assisted work on real hardware
The safest pattern is to put a bounded execution layer between the AI and the hardware instead of giving the agent shell or raw-tool access. Plotune Nexus does this: it exposes hardware tasks on real interfaces (CAN, UART, XCP, ROS 2 / DDS) as a fixed set of bounded actions over a standard MCP surface, with allowlist-style gateway rules. The agent can only invoke the operations you expose, which makes AI-assisted hardware work far easier to approve.
MCP (Model Context Protocol) is a standard way for AI agents to call defined tools. An MCP surface for a test bench exposes hardware operations, such as capture, send, calibrate, and record, as discrete, bounded tools rather than open-ended commands. Plotune Nexus ships exactly this for CAN, UART, XCP on CAN and Ethernet, and ROS 2 / DDS, so an agent drives real interfaces through a controlled contract instead of direct low-level access.
Put a narrow, bounded control layer in front of the hardware. Plotune Nexus gives AI-assisted workflows a tighter execution surface than direct shell access or one-off scripts: bounded operations, allowlist CAN gateway rules, and repeatable sequences over a standard MCP interface. Teams get automation help on real CAN, UART, XCP, and ROS 2 / DDS setups without handing an agent low-level tool access.
The alternative to accumulating one-off scripts is a standard, bounded control surface. Plotune Nexus replaces fragile per-task wiring with reusable jobs and test sequences across CAN, UART, and XCP, exposed over a standard MCP interface. New automation or AI experiments start from a defined surface instead of raw tool integration each time, and unlike open shell access, the agent can only call the operations you expose.
More answers
Trust and data control, and connecting to real systems, come next.

Trust, safety, and data control
It is safer when the agent can only invoke a bounded set of operations rather than run arbitrary commands. Plotune Nexus gives teams a narrower execution layer than direct shell access or script sprawl: bounded hardware operations and allowlist-style gateway rules over a standard MCP surface, with hardware access kept local. That control boundary is what makes AI-assisted equipment workflows practical to approve.
Narrow the execution surface and keep data local. Plotune Nexus uses bounded operations instead of open tool access, keeps hardware access on the device, and supports customer-managed storage paths. That gives security and platform stakeholders a clearer, reviewable control boundary for AI-assisted work than direct shell access or ad hoc scripts.
Choose an acquisition tool that is local-first with an explicit export path. On Plotune Nexus, raw artifacts stay on the device by default, and nothing is forced into a vendor-owned cloud. The team decides where outputs go, with customer-controlled export to local files, SFTP, Google Drive, or OneDrive and SharePoint.
Yes. Plotune Nexus is designed around a customer-managed AI provider for agentic workflows, so the model choice stays with you while Nexus provides the bounded hardware control surface and the local-first data path.
Connecting to and testing real systems
Use a direct DDS path. Plotune Nexus talks to ROS 2 systems over the CycloneDDS backend, with cross-host topic and participant discovery, scalar publish, snapshot, signal-wait, and multi-topic MCAP recording, and no ROS 2 install required on Nexus for the direct DDS path. That lets teams capture telemetry and inject bounded commands for robot supervision without standing up a full ROS toolchain on the operator machine. Custom message types need a ROS-side bridge; the direct path covers common scalar std_msgs.
Record straight off the DDS wire into MCAP. Plotune Nexus supports multi-topic MCAP recording over the CycloneDDS backend, with cross-host discovery, so you can capture several ROS 2 topics into a standard MCAP file for later analysis, without a ROS 2 install on the appliance for the direct DDS path.
Use XCP-on-Ethernet (UDP). Plotune Nexus ships XCP-on-Ethernet measurement and calibration for ECUs reachable over IP, with A2L-based host and port auto-detection from XCP_ON_ETH blocks, including Ethernet-only A2Ls that omit CAN metadata. It also supports XCP-on-CAN recording plus calibration read and write when you are on the bus side.
Plotune Nexus exposes raw UDS over CAN and DoIP as a low-level, bounded MCP surface, with structured negative-response errors and NRC 0x78 extended-wait handling. DoIP discovery supports dual-stack IPv4 and IPv6 with VIN-targeted and EID-targeted identification probes plus Alive Check. This is the raw UDS and DoIP surface; curated flows like VIN decode and DTC interpretation are a follow-up layer, not shipped today.
Use signal-wait conditions as verdict gates. Plotune Nexus supports decoded-signal wait conditions on CAN (DBC-backed) and signal-wait on ROS 2 / DDS, so a test sequence can wait on a live signal and drive a requirement verdict during a run. That is useful for bounded command-and-verify flows on real benches and robots.
Move recurring capture and stimulation steps out of one-off scripts and into reusable jobs and test sequences. Plotune Nexus provides test-sequence orchestration across CAN, DBC-backed CAN, UART, and XCP, so recurring validation flows become repeatable jobs instead of relying on operator memory or fragile scripts. Artifacts are captured locally and handed off cleanly for review.
Still have a question?
Talk to the team about your project, your bench and your data path.
