Co-Founder https://nitter.gq/t.co/KwXjxubObe Advisory Board Cleveland State University Computer Eng. #IIoT #AI #ML #CxO #IndustrialTransformation #ICSCybersecurity #AR
United States
Joined January 2009
- Tweets29.3K
- Following5K
- Followers31.7K
- Likes24.8K
Tag naming is among the first casualties when AI scales beyond one factory. The probability that sensors carry consistent identifiers across 20 sites is low. Partner content with @InfluxDB. #influxdata_iiot
Note: Video generated by AI.
Traditional historians designed for single-site visibility lack the architecture to feed real-time data across a factory network for AI inference and prediction, a gap that only becomes visible when a manufacturer attempts to replicate a working single-site model at enterprise scale.
Merck treats AI agents the same way it treats human employees on documentation: both can produce different outputs when prompted again, but both must leave behind a defensible record at the time of the original decision. An AI that cannot do this fails the same standard a human employee would fail. Partner content with Adlib. #adlib_iiot
The AI notebook concept captures prompt history, retrieval provenance, and model version at decision time. Without that snapshot, any explanation produced later is a new inference, not a record, and ALCOA's Contemporaneous requirement treats the two very differently.
At a plastics factory, connecting machines and having statistical process control running took just one week after deploying a Unified Namespace architecture. Partner content with @HighbyteInc. #highbyte_iiot
Note: Video generated by AI.
Implementation timelines vary based on each facility's existing processes, systems, and documentation. Much of the work involves understanding how equipment and processes operate and organizing that information so data is actually useful to applications.
Lucian Fogoros retweeted
AI could create extraordinary wealth. But telling a generation that abundance is coming while they can’t afford a home or start a family is how you lose their trust.
I’m incredibly optimistic about what we can build. People need a meaningful stake in it.
Applying HAZOP guidewords more, less, none, reverse, late, other than to every AI action in OT reframes how risk is assessed. Partner content with @skkynetinc. #skkynet_ai #everline_ai
Gary Tillery of Skkynet notes that AI-native companies are at the forefront of managing organizations with both human and digital employees, and the broader industry can learn from how they are approaching this operating model rather than waiting for an HR-style policy to fill the gap.
Model accuracy is the wrong metric to bring to a VP of operations because it describes how the algorithm performs, when leadership needs to know how the operation performs. Three better metrics: mean time to explanation, trust-adjusted risk, and adoption rate, all baselined before AI is introduced.
Trust-adjusted risk assigns a dollar value to the outcome if the AI is wrong, including reputation damage, supply chain disruption, regulatory consequences, and legal exposure. Presenting this figure forces leadership to quantify what they are risking by following or ignoring the AI output.