Most Companies Don't Fail at AI – They Fail Before It Even Starts

2 points by kajolshah_bt 6mo ago ↗ HN

I’ve been seeing more teams talk about “adding AI” as a default next step, even when the underlying problem isn’t fully clear.

In practice, the failures I’ve seen rarely come from bad models or tools. They come from skipping basic questions: – What decision or task is being improved? – Are rules still working at their current scale? – Is there real usage, or just expectations?

I put together a simple breakdown of how AI projects tend to succeed when they do, and where teams usually go wrong early on.

Curious how others here decide when AI is worth the added complexity and when it’s better to wait.

2 comments

[ 4.1 ms ] story [ 15.8 ms ] thread

natepoirier 6mo ago ↗

I’ve found that the real failure point is whether the team can explain what decision would actually change if the system worked. If no concrete action shifts, adding AI just adds complexity without leverage.

kajolshah_bt 5mo ago ↗

Strongly agree. In my experience, asking “what concrete action changes if this works?” filters out most premature AI ideas. If no one can point to a changed decision, AI just adds cost and complexity without leverage.

When the answer is vague (“better insights”, “faster responses”), AI tends to add surface polish without leverage. The projects that stick usually start with a very boring question: what specific action will someone take differently tomorrow?