About / The name and the practice

Knowledge,
built over time.

AEON / time
+
NOETIC / understanding

Aeonetic brings together aeon, a long span of time, and noetic, relating to thought and understanding. The name reflects an interest in how knowledge accumulates through learning, reflection, and experience. In practice, that means understanding a system before changing it, questioning assumptions, and learning from what happens when the work meets reality.

An independent practice

The experience behind the practice.

Aeonetic is an independent AI engineering practice founded by Rob Royce. His prior work at NASA's Jet Propulsion Laboratory and Matter spans agents connected to complex systems, AI-assisted software delivery and natural-language product interfaces.

That experience informs the practice: understand the constraints, make the work inspectable and test it in the context where it needs to be useful.

Prior work at Matter Intelligence

AI in the workflow.
AI in the product.

Engineering the way software gets built.

At Matter, Rob built the flight software team's AI-native development harness across Cursor, Claude and Codex, combining code-review agent skills with quality gates for generated code.

5×merge-request throughput
4 days → 5 hoursMean merge-request open-to-close time

Workflow results reported in Rob’s résumé. Mean merge-request open-to-close time: four days to five hours.

Natural language as a product interface.

He also built a geospatial platform that exposed map, search, navigation, layer and legend controls as agent tools, enabling operation through natural language. Rob demonstrated the platform to prospective utility and insurance customers.

ROSA / Prior research and engineering at NASA JPL

Natural language,
connected to robot software.

ROSA connects natural-language requests to tools for ROS1 and ROS2, supporting inspection, diagnosis and operations. Rob is first author of the ROSA paper, developed with collaborators at NASA's Jet Propulsion Laboratory. The public code makes the engineering inspectable.

The paper reports demonstrations with NeBula-Spot in JPL's Mars Yard, EELS in a laboratory, and Carter in simulation. These are small-scale proofs of concept; the paper calls for further study.

  1. 01 / RequestNatural language
  2. 02 / InterfaceROSA + defined tools
  3. 03 / SystemROS1 / ROS2
A simplified architecture

These examples are Rob's prior work at Matter and NASA JPL, with collaborators. They are not Aeonetic client engagements or institutional endorsements.

Start with the problem

Have a system that needs to work differently?

Tell us what needs to change.

Discuss your AI priority