Research reports
A growing monthly library of benchmark readouts, protocol changes, workflow findings, and detailed studies.
Published reports
Using a Decision Model as an Agent's Guardrail, Not Its Judgment
Two simulated control systems, 88 real paid model calls, and one consistent finding: a fast decision model is a real safety win behind a rule, and a measurable safety regression left to decide on its own.
September 2026: Agentic Testing Environments Monthly Benchmark Review
A category-normalized review of 13 models across six agentic engineering workflows, with final validation index, cost, and duration context.
AI Agents vs. Traditional Test Automation: What Should Actually Be Autonomous?
A framework for AI test automation that scopes autonomous testing by how bounded and reviewable an action is, not by how capable the underlying model happens to be.
From Manual Test Benches to Unattended Testing: Building a 24/7 Hardware Test Lab
A practical look at what it takes to turn a manual test bench into an automated test bench that supports unattended testing and remote hardware testing around the clock.
Agentic Testing for ROS 2 & DDS Systems
A walkthrough of how an AI agent runs ROS2 test automation end to end: discovering the DDS graph, commanding and gating on telemetry, recording multi-topic evidence, and producing a verdict.
Agentic ECU Testing: From CAN Signals to Autonomous Test Execution
A look at the decode layer underneath CAN test automation: why DBC-based signal decode, not raw frames, is what makes ECU test automation and autonomous execution trustworthy.
How AI Agents Can Control Real Test Benches with MCP
Why MCP hardware test automation works where raw shell access doesn't: a narrow, typed, authorized set of operations is what makes AI agent hardware control safe to approve.
AI Hardware-in-the-Loop Testing: Giving AI Agents Access to Real Hardware
Why most HIL rigs keep AI agents out entirely, and what a bounded-access architecture for AI HIL testing looks like once it stops being a trust exercise.
What Is Agentic Test & Development? From Test Scripts to Autonomous Engineering
What "agentic testing" actually means, and how agentic test automation differs from a fixed test script by deciding its next bounded action instead of replaying a fixed one.
Why test automation still requires an engineer in the loop
Automation that stops at "run the script" still needs a person standing next to it: a look at where test automation actually breaks down and what closes the gap.
Running unattended ROS 2 hardware tests
Why most ROS 2 hardware testing is bottlenecked on a terminal and a person watching it, and what it takes to make a DDS-based test run unattended.
Automating CAN-based ECU tests with AI agents
Why a CAN or ECU test procedure is a lifecycle, not a single frame send, and how an agent-driven workflow keeps that lifecycle bounded and reviewable.