AI assistant for PI data
People ask questions about the operation in plain language, and get answers that show the tags, assets and time ranges they come from.
The answers are in PI, but only a few people can get them.
Finding the right tag, time range and trend takes skill and time.
Most people never ask, or wait for someone who can.
“What was the mill throughput last night, and why did it drop at 2 am?”
02What you get
-
Plain-language questions
In English or French, about production, equipment and events.
-
Answers with sources
Each answer shows the tags, AF elements and time ranges it used.
-
Read-only
The assistant reads PI data. It never writes or changes anything.
-
PI security respected
People only see the data they are allowed to see in PI.
-
Built on AF
It uses your asset model to understand what each value means.
03How it works
-
01
Check the foundation
Review AF context and data quality; fix what blocks good answers.
-
02
Connect
Set up the assistant with read-only access and your PI security.
-
03
Test with users
Real questions from operations and maintenance, checked against PI.
-
04
Roll out
Open it to more people, with guidance on what to ask.
04Outcomes
-
More people use the data
No need to know tag names or tools.
-
Answers you can check
Every answer points to its source in PI.
-
Faster investigations
From question to trend in one step.
05A good fit when
- Your PI System has good AF context, or you plan to add it.
- Many people need data, but few know the PI tools.
- You want AI that stays within your security rules.
06Common questions
Can it change our data or settings?
No. It is read-only.
Where does our data go?
We agree with you where the assistant runs and which AI service it uses, to match your IT and security policies.
Do we need AF?
It works best with AF. Without context, answers are limited to tag names. Asset contextualization is often the first step.
Start a conversation.
Tell us about your PI System and what you need. We will suggest a practical next step.