6. Fundability

← All research

6.1 Would I invest at pre-seed or seed?

Pre-seed: No. And the founder shouldn't want me to.

The generic version is a feature. QuickBooks has already shipped it, with AI, for English statements (verified 24 Aug 2026). Investors fund platforms, not features that platforms have just absorbed.
The recommended version is a niche. Its realistic ceiling for a small team is in the low millions of ARR. The proven solo outcome is ~$300–500k a year. That's a great business to own and a poor venture return.
The venture-scale neighbors (lender underwriting, enterprise forensic platforms) need enterprise sales teams and compliance, which is the opposite of the operating model that suits this founder.

I'd look at it only if all of these were true:

≥$15k MRR, growing ≥8% a month;
≥40% of revenue from Firm plans;
a clear path into an adjacent enterprise market (for example, law-firm e-discovery vendors reselling it).

Seed: No.

6.2 What a skeptical investor says in the first five minutes

"QuickBooks does this for free now. Why does this company exist?"
"Model cost is a fraction of a cent a page. What stops your price going to zero?"
"Bank Statement Converter and DocuClipper have years of search rankings. How do you get found?"
"Your users have one case a quarter. What's your churn?"
"Valid8 already verifies against source evidence for forensic firms. Why you?"
"What happens when you drop one row in a court exhibit?"
"Why are open banking and bank feeds not going to end the PDF?"
"How big can this get? Show me the path past $5M ARR."
"Where does your test data come from, and do you have consent to keep it?"
"Why this founder?"

Questions 1, 3 and 4 are the ones that matter. The answers:

The reconstruction job QuickBooks doesn't touch: many statements, proof, citations, non-QuickBooks work.
Use-case search plus professional communities.
Annual plans plus case passes.

Question 8 has no good answer, which is why you shouldn't be pitching.

6.3 Three proof points that would change the answer (to "profitable business worth building", not "fundable")

#	Proof point	Threshold	Why it matters
1	Case users pay and stay	≥50 paying accounts, ≥70% month-3 retention, median ≥300 pages a month per account	Proves they're reconstruction users, not one-off converters.
2	Proof is the reason they buy	≥97% first-run balance pass on new uploads, and ≥50% of trial users cite "balance check" or "source links" as the reason they paid	Proves the moat is real, and not just "a cheaper DocuClipper".
3	Organic acquisition works	≥50% of new paid accounts from organic search or referral by month 9, with zero paid ads	Proves the comparable's channel still works for a 2026 entrant.

A strong fourth: a saved-layout library of 300+ US institutions with measured pass rates. It's a real, compounding asset. A buyer (a practice-management or e-discovery vendor) might pay for it one day.

6.4 Bootstrap or raise?

Bootstrap. Clearly.

It's cheap to build and run. The comparable launched for under $100 and runs on ~$500 a month of costs (verified, Feb 2024 interview). Model costs are a fraction of a cent a page.
Revenue arrives early. Firms pay from the first case. You don't need to buy growth to learn whether it works.
Money wouldn't fix the real constraint. The constraint is search position and trust, which compound with time, not spend. Paid ads have already been shown to lose money in this category.
It fits limited hours if designed for it: self-serve, automated layout learning, no sales calls after the first 20 customers, and support capped by the "report a bad row" loop.
Exit option: a profitable, low-support tool with a search position and a layout library is sellable. Micro-SaaS marketplaces sell tools like this for several times annual profit (not verified here). That's an owner's exit, not a venture exit.

Honest bottom line: the version as described isn't worth funding or building. The reconstruction version doesn't need funding. It needs 18 months of steady publishing, obsessive accuracy work and one good design partner.