ConGro AI just got a significant update — the biggest change to how the app works since launch. Here is what is new.
Agent Mode
Give it a whole task and it works autonomously. It plans the steps, executes them, verifies the result against your live model, and reports back. You are no longer driving it one action at a time — you describe the outcome you want, and you watch it get there.
Chat mode is still there for focused, one-at-a-time exchanges. You pick the mode in the composer, per message.
One unified system
No more choosing models or tiers. The picker is gone. Everyone gets full capability on both plans — the plan you are on changes how many credits you get each month, and nothing else.
Chat mode is significantly better
Faster and sharper than it was, and it reads images now. Paste a screenshot, a sketch, or a detail from a drawing set and it can work from that.
Cleaner UI
Fewer toggles and clearer controls. The one that matters most day to day: you can see what it is doing while it works, instead of watching a spinner and hoping.
Leaner under the hood
The internals were re-engineered so more of every credit goes to real work instead of overhead.
Safer by default
It asks before doing anything destructive, and it will not quietly discard your analysis results. If something is about to be overwritten or thrown away, you get asked first.
On credits
An Agent run is billed as a single message and metered per action, with a cap before it stops and checks whether you want it to continue. The pricing page has the full breakdown, including what each action costs in ETABS and Revit.
If you design buildings for a living, put Agent Mode on a real task and tell me where it falls short. That feedback is what shapes the next release.