Several months ago, I joined a WhatsApp community with around 500 people. Once a community becomes large enough, you start noticing certain patterns.

Whenever someone announced a new product idea, the responses were usually the same:

“That sounds amazing.”

“I would definitely use it.”

“Great idea.”

“Can’t wait to try it.”

I do not think people were intentionally being dishonest. Most were simply being kind. Nobody wants to be the person who discourages someone before they have even started.

But that is also the problem. Agreement is easier than challenge, and encouragement can easily be mistaken for validation.

The second moment happened at a local accelerator hub. I met a founder who had hired a developer to build an online education platform for her workplace-harassment awareness programme.

I asked whether she had used ChatGPT to pressure-test the idea before committing to development. She confidently told me that she had.

That conversation stayed with me.

Using an LLM is easy. Knowing what to ask it, which assumptions to challenge, and what important questions you have completely missed is much harder.

That gap between what you do not know you do not know, and the moment you finally recognise the gap and begin investigating it, is what I built IdeaGrit to address.

IdeaGrit helps founders pre-mortem an idea before committing significant time, money, and energy.

Rather than asking an LLM to evaluate an idea from a single open-ended prompt, the product guides it through category-specific validation workflows using nine predefined cards.

I think of it like a detective investigation.

Give a detective one clue, and the entire town may be a suspect. Give them two clues, and half the town may still be under suspicion. Give them several independent clues, and one explanation starts becoming much more likely than the others.

The same principle applies here. Each validation card provides another independent signal. By the time IdeaGrit produces its final assessment, the reasoning is based on multiple perspectives rather than one line of thought.

The goal is not to tell founders whether their idea is “good” or “bad.” It is to expose the assumptions, blind spots, and unanswered questions they should investigate before deciding what to build.

I would love feedback from the Hacker News community, especially on the validation workflow, the usefulness of the output, and where the reasoning still feels weak.

Here is the product https://ideagrit.foundersailab.com