How FuturePath AI reaches every recommendation — in the open

Every score, salary estimate, and cost figure is produced by a deterministic, auditable formula running on public data — not by a language model. The same inputs always produce the same outputs. The entire recommendation engine runs with no AI at all.

AI is optional and strictly limited: it may reword an explanation into plainer language, but it never invents a number, never changes a score, and never makes a guarantee. If we cannot show you the math behind a figure, we do not display the figure.

Each career is scored 0–100 on six independent dimensions, then combined with fixed, published weights. Nothing is hidden or personalized behind the scenes:

The lifestyle you describe (city, housing, family, savings) is first converted into the annual income it realistically requires, using cost-of-living, rent, fuel, tax, childcare, and savings assumptions — each shown as a line item you can inspect. A career then scores well only if it can plausibly fund the life you actually want.

For every career we show a range: the 10th, 25th, 75th and 90th percentile wages of people currently working in that occupation, plus the median. The affordability verdict — whether a path can fund the life you described — is measured against the lowest of these, the 10th percentile.

That figure is a proxy, not a starting salary. No public source publishes verified first-year pay by occupation, so we use the 10th-percentile wage of current workers as a deliberately conservative stand-in. Some genuine new entrants earn less — particularly in fields with apprenticeships, unpaid or low-paid training, commissions, self-employment, military pay scales, or strong seniority effects. Every recommendation states this limitation alongside the number, and we flag occupations where it applies most.

We chose the conservative figure on purpose. Measuring affordability against the median would answer an easier question than the one a family is actually asking, and would make paths look more affordable than they are.

Some occupations appear without a recommended major. That is deliberate. We validate every occupation-to-program pairing, and when a mapping cannot be defended — because the credential does not qualify someone for the occupation, or because the link is unreliable — we show the occupation with no major and say the education pathway needs review, rather than guessing.

Licensed and graduate-entry professions are the most common case: an undergraduate program alone does not make someone a lawyer, pharmacist, physician assistant, physical therapist, or airline pilot. For those, confirm requirements with a counselor or the relevant licensing authority.

Our estimates are built on official public and government data. Every figure on every card is labeled with its source and date, and — critically — whether it is live official data or a clearly-marked seed estimate used when a live connector is unavailable. We never disguise an estimate as a measurement.

See the live status of every connector on the data sources page.

• Local-first by design: your profile, goals, and recommendations are stored in your browser, not on our servers. You can use the full tool without creating an account.
• Minors are protected: users under age 13 require verifiable parental consent, consistent with COPPA and applicable US law.
• We never sell your data or a student's data, and we never share it with third parties for advertising.
• You can export or permanently delete your data at any time — data controls. Full details in our privacy policy.

FuturePath AI is built to support a counselor's work, never to replace it. It is a conversation-starter grounded in real data — a way to help a family explore options and see tradeoffs, then bring an informed discussion to a human advisor.

• No-degree bias: trades, apprenticeships, certifications, community-college transfer, military, and entrepreneurship are weighed on equal footing with four-year degrees, and every recommendation carries a lower-cost alternative.
• It never discourages a student based on grades alone and always offers a concrete improvement path.
• The tone is calm, balanced, and classroom-safe: no hype, no certainty language, no shaming about money.

• Transparent data provenance: every figure traces to a named public source with a retrieval date. Recommendations draw on BLS wages, Census, HUD and EIA; school figures come from College Scorecard. We list only the sources that actually feed a calculation.
• No guarantees of outcomes: all salary, growth, and cost figures are labeled as estimates from public data, never as promises of admission, employment, or income.
• Privacy-regime aware: the platform adapts consent, export, and erasure rules to the applicable regime (US / GDPR) and localizes its disclaimers accordingly.
• Accessible and standards-minded: semantic, keyboard-navigable pages with clear contrast, built to keep improving toward WCAG conformance.

The commitments on this page are not just words. The deterministic engine is protected by an automated test suite of 128 unit tests, run on every change, with an enforced coverage floor on the core logic. Those tests assert the promises themselves — that risks are always surfaced, that a student is never shamed, that entrepreneurship is never oversold, and that likelihoods are always expressed relatively, never as a false exact probability.