Skip to main content

Just a question for my learning curve. This time about AI. It is there for some time now, but hits IFS Cloud massively from IFS25R1.
In this community thread, I read multiple cases such as:
'...creating work orders from scanned documents. This is recognising that a lot of customers employ to third parties to generate reports, go and do inspection work. Those reports are often really, really fantastic sources of great data and often in those reports you'll find things like, we recommend you change this philtre in three months time. Or we recommend you do an inspection in five months time. And quite often that those kind of recommendations get missed. So basically, we've bought a capability where we'll scan those reports in through PDF format. Look those keywords, look for key phrases, and AI will generate a list of recommended work orders to be created. And you can choose, say AI goes ahead and create the work orders for you, or you can just present it as, here's a list of things that we think we found in the report and go validate. Massive time saving, great opportunity to reduce the risk associated with missing those recommendations.'
This is for a number of companies a welcome addition. This is from the software vendor (IFS) so less worry if it works ok. An overseeable investment.
But what if a company requests for an AI solution that is not on the roadmap (yet)?
What steps should the company take to get it up and running?
There is the investment to get AI in IFS up and running.
But from a technical aspect? Should there be an AI expert that prepares a model? Does that mean investing in Python which is commonly used in the AI world?
Or search for an AI solution (CoPilot, ChatGPT, ...) that might fit best and use their API functionality?

Any comment will help me, so thanks for taking the time to include your thoughts.

I noticed an opening to what is in IFS ai (Industrial AI), which gives more info on what might be possible.

Based on this week’s Connect event the intent seems to be to leverage the IFS-delivered AI agentic processes rather than building your own. Per the sessions there are already over 200 processes in 25R1.  My guess is that long term there may be ways to develop/deploy your own agents within IFS but this hasn’t been indicated anywhere that I have seen or heard.

Creating your own model outside of IFS using a different tool would mean that if you also use the IFS-delivered agents you’d presumably end up with a split model in terms of use, and you’d also need to have it learn the IFS data from scratch.

Nick

Makes sense that IFS will come with a number of possible AI solutions.

If in the beginning IFS ai is more or less a closed box for IFS Partners/Customers, we will most likely have to pay and wait for something that is coming our way. The ideas may then proof to be very important, next to discussions with the customer success manager of IFS.

Is this a vendor lock in situation?

Learned something via: https://blog.ifs.com/2025/05/ai-in-action-start-small-scale-fast-with-ifs-ai/ There is the IFS.ai Starter Package for ready to start material and, for those that need complex issues to be solved: Nexus Black (and also more general info: https://www.ifs.com/nexus-black).

Reply


Terms & ConditionsCookie settings