EjectSeat

How EjectSeat works

EjectSeat reads public disclosures and produces a forward-looking layoff risk score. Everything we use is public — no insider data, no scraping private systems. Here's what happens when you search a company.

The data we read

For every search, we pull the most recent filings and public statements from the company and the media covering it. That evidence bundle goes to our analysis model — we do not use social media, anonymous workplace forums, or unverified leaks.

SEC EDGAR
8-K, 10-K, 10-Q, 20-F, 6-K, and NT filings. Full text — not keyword-filtered — so material disclosures in unexpected Item sections aren't missed.
Earnings calls
Recent prepared remarks and Q&A transcripts from Motley Fool. The Q&A is where CEOs often speak off-script about headcount and cost pressure.
Tier A media
Reuters, Bloomberg, Financial Times, Wall Street Journal, Associated Press, MarketWatch.
Tier B media
CNBC, The Information, TechCrunch, Axios, Fortune, Business Insider.
What we exclude
Reddit, Blind, LinkedIn posts, anonymous forums, workplace gossip, unverified tips.

The four risk states

Each company lands in one of four states. The states overlap on purpose — the analysis model can reflect nuance within a state, and the score inside the band conveys how firm the verdict is.

CLEAR
Score 0–35
No material layoff signals. No confirmed announcements. No forward indicators of material restructuring.
WATCH
Score 25–64
Forward indicators present — activist pressure, CEO cost-cutting language, financial stress, sustained losses — but no confirmed layoff announcement.
LIKELY
Score 45–78
Multiple independent corroborating signals point to restructuring, or a confirmed small cut alongside additional pressure. Strong forward probability.
ACTIVE
Score 60–90
A confirmed layoff event in an 8-K/6-K/press release within the last 90 days, a publicly disclosed multi-year transformation programme currently mid-cycle, or a court-filed bankruptcy.

Strict evidence binding

Every factual claim in the analysis — every confirmed event, every forward-looking signal, every headcount estimate — must cite a specific source from the evidence bundle. We require both a reference (the SEC accession number, article URL, or transcript link) and a direct quote.

A post-validator runs after the model responds. It strips out any claim whose source reference doesn't match something actually in the bundle, enforces a confidence floor based on how many independent sources corroborate a finding (three for high confidence, two for medium), and flags results for review when the verdict and evidence disagree.

This means the score is anchored to what the company and credible media have actually said — not to training-data recollection or surface-level pattern matching.

What EjectSeat isn't

Questions or feedback

EjectSeat is operated by Admin-Ace LLC. Reach us at enquiries.talkace@gmail.com — we read every message.