AI Company Signals.
Per-company AI posture. Hiring intent, comp pressure, attrition risk, AI maturity, pay transparency. 488 AI-active companies. Weekly updates. Bloomberg-grade traceability on every signal.
What it is, and why it exists.
The AI Company Signals Engine answers questions that aggregate market data cannot: is this specific company hiring up or winding down their AI org? Are they paying above or below market? Are they losing senior people faster than peers? Have they actually built a serious AI shop, or is it nameplate?
Every AI company is hiring from the same pool of fewer than 100,000 senior engineers. That pool is small enough that company-specific dynamics matter more than aggregate averages. A candidate weighing two offers from the same funding stage in the same city wants to know which company is in growth mode and which is in cost-control mode. A sales team mapping AI accounts wants to know which companies are about to ramp hiring (and will need to buy more compute, data, observability). A comp consultant building a peer set wants to know which competitors are paying above their stage-specific market.
The engine tracks five signals per company: hiring intent, comp pressure, attrition risk, AI maturity, pay transparency posture: and updates them weekly or monthly depending on the underlying data velocity. Every signal carries a signal_strength field (0-1) so customers can filter for high-confidence rows. Every signal traces back to a public source you can cite: job posting URLs, SEC filings, public announcements, GitHub activity, arXiv affiliations.
We only source from public, ethically-collected signals. No leaked databases. No private LinkedIn scraping. No insider information. Customers can use the output in sales-team research, recruiter target lists, candidate decision-making, and investor research without compliance concerns.
The five signals.
Trailing 90-day open AI requisitions, week-over-week growth rate, and net change in posting volume. Surfaces which companies are hiring up vs winding down.
Trailing 60-day median offer (where disclosed) vs market median for the role, city, and seniority. Surfaces companies offering above or below their stage-specific market.
Trailing 90-day departure rate at the AI / ML org level, normalized against the role-specific market baseline. Surfaces concentration of departures in critical roles.
Depth and breadth of AI talent: total AI engineers / ML researchers / AI infra engineers, seniority distribution, public technical output. Surfaces serious AI shops vs companies in name only.
What fraction of the company's open AI requisitions disclose salary ranges, and whether those ranges match observed offers. Distinguishes companies that disclose honestly from those that disclose to comply but pay outside the range.
Use cases.
Two offers, same stage, same city, similar total comp. Which company has positive hiring intent and stable attrition? Which one is paying at the 75th percentile vs the 50th? Which has actually built deep AI talent? The signals provide the answers candidates cannot get from headline numbers alone.
AI infra vendors (compute, observability, evals, data) use hiring intent + AI maturity signals to prioritize their account list. A company hiring 40+ AI engineers per quarter with growing AI maturity is a prime account for infra. A company shrinking AI hiring is likely not.
Comp consultants identify the comp pressure of every named peer. A company at the 75th percentile comp-pressure mark is paying above market; one at the 25th is paying below. Combined with the Comp by Stage engine, the peer set becomes calibrated to the stage-specific market and the company-specific premium / discount.
AI-focused investors monitor portfolio companies for hiring intent (a leading indicator of growth or contraction) and attrition (a leading indicator of cultural health). Aggregated across the portfolio, the signals form a leading-indicator dashboard that complements financial reporting.
Sample response.
A typical GET /api/v1/intelligence/salaries/companies/[slug]/signals response. All five signals plus the standard envelope.
Methodology.
Each signal has its own pipeline. Hiring intent reads from the same job-posting aggregation that powers the Hiring Velocity engine, filtered to AI-tagged requisitions at the company level. Comp pressure joins the company's offer disclosures (from postings + community submissions) against the stage-specific market band from Comp by Stage. Attrition risk reads from public position-change signals (LinkedIn departure announcements, H-1B LCA absence patterns). AI maturity reads from public technical output (GitHub activity, arXiv affiliations, conference papers). Pay transparency reads from the company's posted ranges and compares against observed offers.
The signal_strength field is the ratio of distinct signals that contributed to the underlying observation relative to a calibrated baseline. A hiring intent signal with 47 distinct postings is stronger than one with 4. An AI maturity signal backed by 8 conference papers and a public GitHub presence is stronger than one with only headcount estimates. Customers should filter on signal_strength >= 0.7 for production decisions.
Every signal's lineage is available via GET /api/v1/intelligence/lineage/[data_point_id]. The lineage returns the URLs, filings, postings, and public records that fed into the signal, with fetch timestamps and reconciliation rules. Customers can audit any signal back to its raw source.
Honest limitations.
- Beta status. Some signals are stronger for public companies than for private. Customers should filter on
signal_strengthfor production decisions. - Lag. Hiring intent updates weekly; AI maturity updates monthly. Signals are leading indicators, not real-time. A company that just paused hiring may take a week to register.
- Coverage. Only 488 AI-active companies are tracked as of May 2026. Smaller companies appear when they cross the 25-AI-engineer threshold. Companies that ban public job postings have weaker hiring-intent signals.
- Public sources only. We do not purchase, license, or scrape private databases. The signal set is what is publicly observable. If a company keeps everything private, the engine returns sparse signals with low strength rather than fabricating data.
Pricing and access.
Company signal queries require the Scale tier and the company_data:read scope. Scale tier includes the bulk endpoints and the realtime change streams that surface signals updates as they happen. Full pricing at /orbyt-intelligence/pricing.
See also.
Methodology version 2026.2-signals-beta. Last updated May 2026. Hiring intent, comp pressure, attrition risk update weekly; AI maturity, pay transparency update monthly. Engine in beta: filter on signal_strength >= 0.7 for production decisions.