Nirmal Raj - June 8, 2026

Quick or Deep: Giving You Control Over How BlueFunda-AI Thinks

Not every question deserves the same amount of thought. “What’s the syntax for this annotation?” needs a fast, confident answer. “Why is this change going to break three other things downstream?” needs the assistant to actually sit with the problem for a moment. Until now, BlueFunda-AI handled every question the same way — there was no way to tell it which kind of question you were asking.

That changes with this update, alongside a smaller but important fix to a safety limit that keeps one of ABAPer’s review features running smoothly.

Tell BlueFunda-AI How Hard to Think

You can now choose how BlueFunda-AI approaches your question, and you can switch between these whenever you like — right in the middle of a conversation if you want to:

Auto is what you get out of the box — you don’t have to think about it. But if you know a question is simple, switching to Quick gets you out the door faster. And if you’re working through something genuinely tricky, switching to Deep tells the assistant to slow down and reason it through properly. You’re free to flip between all three at any point, on any message, depending on what you’re asking.

Here’s the slightly embarrassing part of the story: Quick and Deep already existed in the chat interface, but choosing them wasn’t actually doing anything. Your selection was getting lost along the way before it ever reached the part of the system responsible for picking the right approach. So whether you picked Quick or Deep, you ended up with the same response either way — quietly falling back to Auto behavior regardless of what you chose.

That’s fixed now. Your choice travels all the way through, end to end, so picking Deep on a hard question genuinely changes how the assistant handles it — rather than being a setting that looks like it does something but doesn’t.

There’s a second piece to this: if you’ve explicitly chosen which model you want to talk to, that choice now takes priority. The thinking-mode selection steps aside and lets your model choice stand, instead of the two settings quietly fighting each other in the background.

It’s a small change to describe, but it’s the difference between a setting that exists in the interface and a setting that actually changes your experience.

Making Sure the 150-Line Limit Actually Holds

ABAPer can analyze ABAP code and tell you what needs to change to get it ready for S/4HANA. For larger pieces of code, that kind of deep analysis is better handled through the File Analyzer or Batch Processing tools, where it can get the attention it needs. For that reason, there’s a limit: code submitted directly in chat for this kind of review is capped at 150 lines, and anything longer gets redirected to the right tool instead.

This week’s work was about making sure that limit actually applies in every situation — because it turned out there were a couple of ways around it that nobody had intended.

One way around it: if you referred to a piece of code by name instead of pasting it directly, the limit didn’t apply at all, no matter how long the code actually was.

Another way around it: a separate path through the system — the one used when you bring your own tools into the conversation — skipped the check completely. Code of any length could go through there without the 150-line rule ever being applied.

Both of those gaps are closed now. No matter how you submit code for this kind of review — pasted directly, referenced by name, or through a different tool path — the same 150-line check applies, and the same clear message points you to File Analyzer or Batch Processing when your code is longer than that. We also added detailed internal logging so the team can see exactly when and why the limit kicks in, which makes it much easier to confirm the rule is working the way it’s supposed to, rather than hoping it is.

A limit that only sometimes applies is worse than no limit at all — it gives a false sense of consistency. Closing these gaps means the rule now means what it says, everywhere.

A Quieter Fix Worth Mentioning

One more change is worth a brief mention even though you’ll never see it directly: a connection check inside one of our backend services had a fallback that, in certain situations, would skip verifying who it was talking to. That fallback has been removed, and the service now always verifies the connection properly. It’s the kind of fix you hope was never actually a problem in practice — and the kind you’re glad to have closed regardless.

Why This Matters

None of this is a flashy new feature you’ll see announced with fireworks. But together, these changes are the difference between a product that appears to respect your choices and limits, and one that actually does — consistently, every time, no matter which path your request takes through the system.

If you’ve switched to Deep before and weren’t sure it changed anything — it does now. Leave it on Auto for everyday questions, reach for Quick when you’re in a hurry, and switch to Deep when a question deserves more than a fast answer. Try it out and let us know what you think.