Hacker News Reader

Comment by alexpotato | Hacker News Reader

Having worked at some giant firms that heavily depend on SAP, I would think their position as "system of record for procurement/production" would get stronger with AI.

e.g. AI + a good database of what your company does seems like an excellent combination.

View on HN

Or AI can be a threat: If AI can spit out good enough special purpose tools a lot of the SAP adaption business goes away. If AI then also assists in data migration the moat goes away.

EAI has existed for twenty years...Tibco, webMethods, IBM AppConnect...SAP is still what it is. How is AI going to make a dent on this?

I've been impressed so far with Snowflake's offerings of "AI + a good database"

The same argument can be made about snowflakes and other providers that handle the data layer. If they can leverage AI in a safe contained environment they would become the clear winner

LOL

Although they may internally use LLM, I wanna say said they were not going to allow their core offering to be integrated with LLM

Although maybe they reversed course?

Would a technology famous for hallucinating data often enough, and in many context unable to guarantee it won't assume and hallucinate when it feels "confident enough" despite a robust system prompt, be really considered for a "system of record for procurement/production" usage?

(yes yes RAG exists)

In the ERP space, AI is being used mostly for repetitive tasks, documentation and forecasting-type activities or where a non-deterministic answer might be useful. Scanning in documents and using OCR to create an order, asking what page I use to do X, historical demand modelling and constraint scheduling. Also have seen it used to do order flow type stuff like suggesting products to sell if they buy ABC. I'm not really seeing it being used to do heavy analysis and they've generally been careful not to put it in places where a hallucination could result in a bad business decision.

> Scanning in documents and using OCR to create an order,

I remember this being done more than 10 years ago.

The other uses cases you describe also were being done. Of course with AI maybe it’s become easier and cheaper?

Yes but it generally expected documents to be in a certain layout and the user had to fill in a lot of the gaps. It's more able to identify elements on an incoming PO, match a supplier invoice, or create a receipt based on a packing list with inference.

EDIT: The other activities, yes I think it made it cheaper and easier. They both required human intervention to make an accurate forecast or execution plan. Forecasting requires humans to do things like apply smoothing factors, identify abormal demand over huge datasets, or identify seasonal demand cycles, etc. Constraint scheduling requires a human to intervene in dispatching and resource allocation as well. AI can do 95% of the work with a proposal, or even control the process. Supposedly much better than historical algorithms. I haven't implemented this part of it, but I've seen it done.

The documentation part is much more robust. You can ask it complex questions, like "How do I setup a phantom blow-through part in a MS level 1 BoM?" and it walks you through the entire thing. With the sales suggestions, it can identify things like "The customer is buying spaghetti and pancetta" AI: "They are probably making carbonara, how about some garlic bread?"

Take your favorite OCR software.

Then take a photo of a document and throw it into ChatGPT, or Gemini, along with the word "Transcribe".

ChatGPT wins, by a landslide. BUT AI's advantage doesn't stop there.

"Take this picture of some idiot filling in form 49, I've also attached the PDF, fill in the PDF fields, provide a database record according to the schema attached and flag if there are any obvious problems with the entries".

That works too. "Produce a latex document of this kid's math homework and flag any problems" - works. "Produce a MS word document of this letter" - works. "Read this bill and produce a JSON version following the schema from this example" ... and so on and so forth.

More than that, it is starting to work pretty well with Gemma 31B local model (will still do 30 document analyses on cpu only, at Q4 on a DDR4 or higher machine. Yes it's mostly memory speed that matters) at this point. I mean, that's GPT 5.0 or so quality (95% correct with the occasional problem), but hey, Qwen 3.8 27B may be coming out next week ...

30 document analysis in parallel

Actually 30 is sequential on an hosted Xeon server. No GPU, about 2 minutes for a ~2000 token answer. I should try parallel since you're quite right, I should be able to do faster that way.

> forecasting-type activities

AI/ML yes, LLM's no (unless someone is naively using LLM's for forecasting)

Reply on news.ycombinator.com