Every night the Origination Engine reads the SEC filings of 674 US-listed companies in Life Sciences, Technology and Natural Resources with a public float (the market value of shares held by outside investors) of $15M to $400M. It ranks which ones are likely to need new capital soon, shows why with a link to each filing, and emails you a one-page summary on the top ten at 5:00 AM US Eastern Time.
$1,500 a month after the 14-day trial. A card is required at signup; cancel before day 14 and you pay nothing. Pricing details
Public SEC filings only. No paid data, no paid experts, no inside information. New to the terms? Plain definitions.
The Problem
You already know what a company worth financing looks like. The hard part is reading enough filings to find it. The engine's universe is every US-listed company in Life Sciences, Technology and Natural Resources with a public float between $15M and $400M, which corresponds to a company value on the stock market (market cap) of roughly $20M to $500M: 674 companies at the last rebuild. Reading one quarterly report (a 10-Q) properly takes half an hour, and each company files several a year. One more analyst does not close that gap.
The Four-Pillar Engine
Each stage is independently checkable, and the output of each one is visible in the brief you receive.
Do not rely on financial ratios that every data terminal can also sort. The engine watches what issuers actually file, every night, and surfaces the events that mean a company must raise money: a stock exchange notice (a “listing deficiency”) that borrowing cannot fix, a warning that the company may not be able to keep operating (a “going concern” warning), a shelf registration too small to carry the raise, or a convertible note coming due.
Every name arrives with its arithmetic shown. Six weighted components out of 100, each one printed with the evidence that earned it, so you can disagree with a specific line rather than with a black box. A name surfaces only at 45 or above.
A small company’s share sale is priced according to how much ordinary investors want the shares. If current shareholders already expect new shares to be issued (which lowers the value of the shares they hold), the company can get different terms. A paid, coordinated promotion of a stock that is about to raise money is a warning sign, not a good sign. Every company the engine surfaces gets one reading of public online discussion, reported with the posts it came from.
Uncalibrated This pillar is new and is not part of the backtest. Every figure on the evidence page comes from the deterministic layer alone.
A ranked list on its own is a screen, and you can already build a screen. What makes this useful is that each name arrives with the work already done: cash and runway, cash spent over the past year, how much the company can still sell under its shelf, its convertible notes and their terms, earlier financings and the banks that arranged them, and the specific filing event that surfaced it today.
The Difference: Filings vs. Screens
Returns every name that shares an attribute. Cannot tell you that a pay committee just linked a bonus to a decision date, or that a program for selling shares at market price was cut to zero yesterday. Those live in prose, in one filing, on one day.
Arrives with the specific event that surfaced it, the arithmetic that makes it urgent, and a link to the document. The engine reads the same filings you would. It just reads all of them, every night.
The system ranks and explains. A person decides. Nothing in this engine buys or sells, contacts a company, sets the size of a trade, or recommends an action.
Compliance is built in.
Every input is a public SEC filing, a public exchange disclosure, or a public post. Nothing in this stack touches material non-public information (inside information), paid expert consultants, or company-identifiable alternative data. No credit card spending data, no location data, no scraping of private systems, and no paid data feed of any kind. So there is no question about information barriers (the rules that keep teams inside a bank apart), and nothing limits what you can do with a company after the engine surfaces it.
vs. Other Tools
Data terminals are priced for large institutions and assume you already know which company you are looking at. Databases record what already happened. Screeners filter attributes. None of them read tonight's filings and tell you which issuer has to raise within weeks.
| Capability | The Origination Engine | Terminals Bloomberg, CapIQ |
PIPE Databases PlacementTracker |
Screeners Finviz, Koyfin |
SEC Resellers sec-api, Calcbench |
|---|---|---|---|---|---|
| Nightly filing-event triggers, ranked | ✓ | ✗ | ✗ | ✗ | Raw feed |
| Baby shelf limit calculated (Form S-3, I.B.6) | ✓ | Manual | ✗ | ✗ | ✗ |
| Runway from XBRL numbers, old data marked | ✓ | Manual | ✗ | Partial | Raw feed |
| One-page dossier, every number linked | ✓ | ✗ | ✗ | ✗ | ✗ |
| Contacts copied exactly from filings, never guessed | ✓ | Directory | ✗ | ✗ | ✗ |
| Public discussion read on each company | ✓ | Separate product | ✗ | ✗ | ✗ |
| Published scoring rules you can keep | ✓ | Black box | n/a | You build it | n/a |
| Lead-time backtest you can check yourself | ✓ | ✗ | ✗ | ✗ | ✗ |
| Priced for a small firm, such as a family office | ✓ | Per seat | Subscription | ✓ | Metered |
The bottom line: terminals cost six figures and assume the name is already in front of you. Databases tell you what closed last quarter. Screeners only filter by company attributes. The Origination Engine is built for the investor or banker who has to find the company first, and who wants to check every number against the filing before making the call.
Plain Definitions
Start With the Measurement
I have not seen your names yet, and anyone who gives you a number before seeing them is trying to sell you something. What I will commit to is the measurement: a rubric written from thirty to fifty names you have actually financed, an honest retrospective, and lead-time numbers you can check against filings yourself. If the retrospective is weak, I will tell you first.
$1,500 a month after the trial, card required, cancel before day 14 and pay nothing. Want proof first? See the twelve-month backtest.
Or call +1 (775) 252-8333 (a US number), at any hour. An AI assistant that speaks for Ariel Agor, who built the engine, answers questions and can set up the trial on the call.
The brief is the email subscribers receive at 5:00 AM US Eastern Time: a ranked list of small public companies likely to raise capital soon, with a one-page summary on each. Tell us what you invest in and we will email you one sample, free. A person follows up the same day.