CareerBloom uses AI in several places, and people are right to ask what each tool actually does. This page is a plain-English explainer for every AI surface you'll encounter — what it sees, what it can’t see, and how to get value from it.
What every Bloom AI output carries
Every AI output on Bloom is tagged with one of three confidence levels, so you know how much weight to put on it:
Data-groundedData-grounded. Pulled from verified CareerBloom data — hires, salary confirmations, Hiring Pulse responses. We show the sample size whenever it matters.
Pattern-basedPattern-based. Derived from patterns in job descriptions and profiles. Useful as a heuristic, not a prediction.
DirectionalDirectional. Informed guidance — use as a starting point, not as a final answer.
Role fit / match score
Pattern-based
A 0–100 score for how well your visible experience aligns with a job's stated requirements. Computed by comparing skills, seniority signals, and language in your profile against the job description. We never claim to predict whether you'll get the role — only how the words on both sides overlap.
What it can’t see
How you'll perform in an interview, or whether you and the hiring manager will click.
Skills you have but haven't described in your profile or CV.
Whether the job description accurately reflects the actual role (we surface this separately with Ghost Detector).
Hiring Quality Index (HQI)
Data-grounded
An employer's HQI is built from Hiring Pulse — short, structured feedback from candidates who actually went through that company's process. It covers response rate, time-to-response, job-description clarity, and rejection communication. Each metric is a percentage of real responses.
What it can’t see
Show anything until an employer has at least 5 Hiring Pulse responses. Below that, we hide HQI entirely — small samples are misleading.
Reveal individual candidate feedback. We only show aggregate percentages, never names or quotes.
Tell you whether a specific recruiter at the company is fast or slow — HQI is a company-level summary.
Bridge This Gap
Pattern-based
When a job lists a requirement you don't obviously meet, Bridge searches your CareerBloom profile for adjacent experience and suggests honest framing — never claiming a skill you haven't described. Every suggestion comes with the exact profile entry it was drawn from.
What it can’t see
Invent experience. If no anchor exists, we say so explicitly and recommend you prepare a genuine example instead.
Verify the framing will land in interview. The framing is plausible because the experience is real — but how you tell the story is your part.
See information outside your profile. Skills that exist in your head but not on the page are invisible to Bridge.
Ghost Detector
Pattern-based
Some job listings stay live long after the role is filled, paused, or quietly cancelled. Ghost Detector looks at signals — listing age, repost history, employer hiring activity — and flags listings that show patterns of inactivity. It's a heads-up, not a verdict.
What it can’t see
Verify whether a listing is genuinely closed. We can see signals, not the recruiter's calendar.
Distinguish 'slow-moving role' from 'ghost role'. Some legitimate processes simply move slowly.
Account for internal hires (where the listing stayed live but the role is gone). This is a known blind spot.
Salary data on listings
Data-grounded
When CareerBloom has enough verified salary data points for a role and region, we show a percentile range. The data comes from candidates who voluntarily shared their offer numbers after acceptance — never scraped, never inferred.
What it can’t see
Show anything below a minimum sample size. If we don't have enough data, we show nothing — not a guess.
Adjust for individual context: equity, signing bonus, hidden allowances, cost-of-living differences within a city.
Tell you what to ask for. Markets are noisy; your number depends on your leverage as much as the median.
Career trajectory / What Happened Next
Directional
We look at where people in a similar role tend to move over the following 3–7 years, drawing on aggregate career-move data (BLS, O*NET, and — where the sample is large enough — CareerBloom's own verified hire history). We show patterns, not predictions.
What it can’t see
Predict your career. We can show common moves; whether you make them depends on you, the market, and a hundred unknowables.
Tell you whether a specific role will lead somewhere — only what others in similar roles have tended to do.
Account for shifts in your industry that haven't shown up in historical data yet (AI breaking out, regulation tightening, etc.).
Our commitment
When we are uncertain, we say so. When data is limited, we show the sample size. We would rather give you accurate context for a rough estimate than false confidence in a precise one.
If an AI output on the site ever feels misleading, that’s a bug — not a feature. Let us know.
