Methodology

Built from public sources. Labeled by confidence.

OnePersonAI is a structured research database, not a rumor board. We turn public information into operating-model analysis while keeping uncertain numbers visibly uncertain.

A

Public sources first

We research official websites, pricing pages, founder posts, interviews, podcasts, public AMAs, product pages, and trusted third-party profiles.

B

Facts and analysis stay separate

Every case separates public facts, source-labeled claims, OnePersonAI analysis, and replication suggestions so readers can see what is evidence and what is interpretation.

C

No unsupported numbers

Revenue, users, team size, and acquisition numbers are labeled by confidence. If a number is not public, we say so. If it is third-party reported, we label it.

D

AI assists the workflow

AI helps us organize research and create first-pass structure, but humans decide what gets published and how claims are labeled.

Confidence labels

Source confidence.

Source confidence describes how directly public material supports the case file.

Label
Meaning
Typical evidence
A
Directly supported by official pages, pricing pages, product docs, founder posts or interviews.
Strong enough to cite as public fact.
A/B
Mostly primary-source backed, with some interpretation or older source material.
Useful with context.
B
Supported by public material, but some details depend on interviews, third-party summaries or inference.
Use carefully.
C/D
Thin, old or indirect public support.
Comparison only; do not overuse.
Revenue confidence

Revenue is not treated as audited unless it is.

Revenue confidence describes whether pricing, customer, revenue or transaction claims can be used as evidence.

Label
How to read it
What we avoid
High
Official pricing, transaction surfaces, public sale data or direct founder disclosure support the claim.
We still do not turn it into audited financials.
Medium
Useful public signals exist, but exact revenue should remain qualified.
We avoid precise ARR/MRR claims without source strength.
Low
There are product or audience signals but little reliable revenue proof.
We focus on monetization logic instead of numbers.
Unknown
No dependable revenue evidence is available.
We do not invent numbers.
AI usage

AI-native, AI-assisted and AI-era are different labels.

The label describes the role AI plays in the operating model, not whether the founder is fashionable.

Label
Meaning
Example use
AI-native product
AI output or AI workflow is central to what the customer buys.
Generated assets, AI software, AI agents or AI-powered product outcomes.
AI-assisted operations
AI helps research, production, support, editing or fulfillment.
Lower headcount or faster publishing without AI being the product.
AI-focused media
The business sells attention or insight around the AI market.
Newsletters, directories and analysis products.
AI-era reference model
Not necessarily AI-native, but useful for low-headcount internet business design.
Paid databases, SEO pages, newsletters, templates and SaaS.
Fit score

What the 0-100 score means.

The score is an editorial usefulness score for OnePersonAI readers, not a revenue ranking, investment rating or claim that one company is objectively better than another.

Factor
Question
Why it matters
Observable distribution
Can we see where attention or demand comes from?
Higher when the case teaches a repeatable operating pattern without overclaiming private data.
Concrete monetization
Is there a clear paid unit, pricing path, or source-labeled revenue model?
Higher when the case teaches a repeatable operating pattern without overclaiming private data.
Low-headcount fit
Can the model plausibly be operated by a solo founder or tiny team?
Higher when the case teaches a repeatable operating pattern without overclaiming private data.
AI-era relevance
Is AI the product, an operational lever, a media angle, or only context?
Higher when the case teaches a repeatable operating pattern without overclaiming private data.
Evidence depth
Are there enough public sources to study the operating pattern responsibly?
Higher when the case teaches a repeatable operating pattern without overclaiming private data.
Review and correction policy

We do not turn uncertainty into fake precision.

If a number is not public, we say so.

If a number is third-party reported, we label it.

If a number is founder-disclosed, we label it.

If a founder, reader or source shows a better correction, we update the case and keep the confidence label aligned with the strongest available source.

The useful product is not the illusion of perfect data. It is a clean map of what is known, what is claimed, and what can be copied.