Trying to set realistic expectations before we commit to anything.
We've started reaching out to a few vendors for a custom AI development company to build a fraud detection tool, and the timelines we're getting back vary wildly. One team said six weeks, another said four months for what sounds like a similar scope. Hard to tell if one is rushing or the other is padding.
First time doing this so we have no internal benchmark to compare against.
Trying to figure out what's realistic:
- Whether a wide timeline gap usually means one team skipped a proper discovery phase
- How much of the delay typically comes from data prep versus actual model building
If you've gone through something similar, how did your real timeline compare to what was originally promised?