Facebook
Google
GitHub
Linkedin
Hello everyone,
I’m the founder of PramaanOne. We’re building Bhairav, an AI system focused on reasoning over industrial and physical-world data.
We’re particularly interested in a simple but important question:
When an AI receives an industrial measurement, how much can it legitimately conclude from that measurement alone?
For example, consider a 3.2 mA reading on a nominal 4–20 mA loop. Depending on the instrumentation, configuration, diagnostics, and available context, there may be multiple possible interpretations. We want to test whether an AI system can recognize that uncertainty rather than confidently choosing an unsupported explanation.
We’re currently building a 100-case industrial reasoning benchmark covering instrumentation, sensors, process data, SCADA, mass balance, environmental measurements, and other physical-world scenarios.
I’ve put three initial cases into a live sandbox:
https://bhairav.pramaanone.com
I’m not posting this to sell software. I’m looking for experienced engineers to try to break the system.
If you have an instrumentation edge case, contradictory reading, sensor failure scenario, process anomaly, or other situation that could expose an incorrect assumption, please test it and let me know where the reasoning fails.
The objective is straightforward: find the cases where the system gets the physics or evidence wrong, document them, and improve it before wider deployment.
Technical criticism is very welcome.
I’m the founder of PramaanOne. We’re building Bhairav, an AI system focused on reasoning over industrial and physical-world data.
We’re particularly interested in a simple but important question:
When an AI receives an industrial measurement, how much can it legitimately conclude from that measurement alone?
For example, consider a 3.2 mA reading on a nominal 4–20 mA loop. Depending on the instrumentation, configuration, diagnostics, and available context, there may be multiple possible interpretations. We want to test whether an AI system can recognize that uncertainty rather than confidently choosing an unsupported explanation.
We’re currently building a 100-case industrial reasoning benchmark covering instrumentation, sensors, process data, SCADA, mass balance, environmental measurements, and other physical-world scenarios.
I’ve put three initial cases into a live sandbox:
https://bhairav.pramaanone.com
I’m not posting this to sell software. I’m looking for experienced engineers to try to break the system.
If you have an instrumentation edge case, contradictory reading, sensor failure scenario, process anomaly, or other situation that could expose an incorrect assumption, please test it and let me know where the reasoning fails.
The objective is straightforward: find the cases where the system gets the physics or evidence wrong, document them, and improve it before wider deployment.
Technical criticism is very welcome.
