Understanding the key needs in investing
Investing is one of the clearest examples of where AI is both incredibly promising and incredibly dangerous.
On the one hand, investment teams deal with large volumes of data: models, PDFs, reports, operating metrics, comps, covenants, market research, diligence materials, portfolio updates, and internal systems.
On the other hand, this is not a space where “mostly right” is acceptable.
A single wrong number, a missed assumption, a broken formula, or a hallucinated source can change a decision when millions of dollars are on the line.
That’s the tension we’ve been obsessed with as we've built Leni.
Everyone wants AI to save time for analysts, associates, asset managers, lenders, and operators. But the real question is not just:
“Can AI generate an answer?”
It’s:
“Can AI produce work that a serious investment professional can actually trust?”
For us, that means AI needs to be able to:
Work across messy real-world files and systems
Understand spreadsheets, reports, and source data
Show where every answer came from
Validate calculations and assumptions
Avoid hallucinating when the answer is not available
Deliver finished work, not just conversational output
We’re launching Leni on Product Hunt today after spending the last few years working closely with investors, operators, lenders, and asset managers on exactly this problem.
I’d love to hear from the Product Hunt community:
Where do you think AI will have the biggest impact in investing?
Is it diligence? Research? Portfolio monitoring? Reporting? Back-office workflows? Underwriting? Something else entirely?
And more importantly, what would it take for you to actually trust AI in a high-stakes financial workflow?
Replies
@arunabh_dastidar Congratulations on the launch!
I’ve worked on similar approaches through AI in the past and I can fully support the point that AI investment output that looks good at first glance often only does so superficially.
The dangers are real. Would you commit your capital on a whim because a friend or colleague had a feeling something might take off? That’s effectively what acting on unverified AI output amounts to.
That said, working across large, varied data sets is probably the single biggest advantage AI can bring to investing. For humans, being right in genuinely complex situations takes either deep experience or luck. AI can shift that equation by processing far more data than any analyst could and turning it into a decision-making input for the human — not a replacement for the judgment, but a serious upgrade to what feeds it.
But trust is the whole game. That’s why I’d want to run Leni against some of my own past cases to see where it actually differs, and what it does better than the workflow I already know.
On that: when your output looks right but rests on a wrong assumption or a misread source, what specifically catches it before it reaches the user?
Is it the citation trail, a separate validation pass, or something else?
Leni
@michael_zorez That's the key: the math models and verification layers we built don't allow any wrong assumptions or misread source creep into the answers. We tie the output train, verifiability, and confidence scores to ensure they're not off at all.