Bernoulli — Industries

Bernoulli, applied. Real lenses across five industries.

A lens is a question the model answers over your input in one forward pass — binary, rating, or choice — with calibrated confidence you can route on. Below are twenty-five lenses from five starter packs that ship with the app at app.bernoulli.live: Salesforce, Terms of Service, Financials, Software Contracts, Shopify. Pick a pack, point it at a Chrome tab, a CRM record, an uploaded PDF, or your own feed — or author new lenses of your own.

Jump to Salesforce Terms of Service Financials Software Contracts Shopify
Salesforce · CRM

Five lenses that an SFDC admin can ship as fields and routing rules.

Lead quality, intent, deal health, churn risk, and tone — the signals that today either live in rep intuition or get filled in late when it's too late to react. Each one runs on the record's own text (lead description, case thread, opp notes) and emits a value you can route on, color on, or escalate on.

LeadChoice
01

Real Buyer Check

Classifies an inbound lead as buyer, spam, student, vendor, or wrong-fit.
buyerspamstudentvendorwrong_fit
Why it matters. Kills junk at the top of the funnel before an SDR wastes a dial. Instantly relatable — every SFDC admin has felt this pain.
LeadRating 1–5
02

Buying Intent

Scores how strong the lead's intent is based on available signals.
Why it matters. A graded signal (not just hot/cold) that routing rules and SDR queues can sort by. Pairs well with #1 — after you throw out spam, rank what's left.
OpportunityBinary
03

Economic Buyer Engaged

Is the actual budget-holder engaged in the deal, based on the opp state and recent activity?
Why it matters. MEDDIC's single most-leaked signal. Pipeline looks healthy until the quarter closes and half the deals turn out to have been multi-threaded with the wrong person. Catches that earlier.
CaseBinaryRisk signal
04

Churn Risk

Does the case language suggest meaningful churn risk — cancel threats, competitor mentions, "last chance" phrasing, demands to escalate?
Why it matters. Routes at-risk accounts to CS leadership in minutes instead of after the renewal call. High-stakes signal that justifies the plugin on its own.
CaseChoice (5-level)
05

Customer Sentiment

Classifies the customer's tone as calm, mildly frustrated, frustrated, angry, or hostile.
calmmildly_frustratedfrustratedangryhostile
Why it matters. The most visible lens — a 5-level color-coded badge on every case view. Great for product demos because the scale shows up well in a UI screenshot.
Terms of Service · Consumer rights

Five lenses that turn a 12,000-word ToS into a one-glance verdict.

The questions nobody actually reads the ToS to find out — and the model answers them in a single pass. Four are binary (fast red/green badges you can surface in a Chrome extension side-panel); one is a 1-to-5 rating that gives the whole document a fairness score to anchor the panel.

Data & PrivacyBinary
01

Can they sell / share my data?

Does this document allow the company to sell or share users' personal data with third parties for advertising, marketing, or commercial purposes?
Why it matters. The most-asked ToS question of the last five years. Answering it with a calibrated probability (not just a keyword search for "sell") is the whole value prop.
Content & AIBinary
02

Can they train AI on my content?

Does this document allow user content or user data to be used to train, fine-tune, or improve the company's AI or machine-learning models?
Why it matters. New question class post-2023; existing ToS summarizers miss it because it's phrased six different ways across providers. The semantic read gets it right.
Money & SubscriptionsBinary
03

Does the free trial auto-charge?

Does this document indicate that a free trial automatically converts to a paid subscription unless the user explicitly cancels?
Why it matters. The single biggest source of "I didn't know I was being billed" complaints. One lens on checkout flows catches it before signup.
Legal RightsBinary
04

Forced arbitration?

Does this document require users to resolve disputes through binding arbitration, and does it waive the right to sue in court?
Why it matters. Opt-out windows for arbitration clauses are often 30 days. The lens flags it in time to opt out, not six months later during a dispute.
SummaryRating 1–5
05

Fairness to the user

On a scale of 1 (one-sided — all rights to the company, obligations to the user) to 5 (balanced — reciprocal rights and obligations), rate how fair this document is to the user.
Why it matters. The whole-document summary row. Pairs with the four binary lenses above to give a one-screen verdict: here's the score, here are the four specifics that drove it.
Financials · MD&A reads

Five lenses that score an earnings press release before the market opens.

Not "summarize the quarter" — specific, decision-shaped reads on management disclosure quality. Did they attribute revenue change to volume or price? Was the margin story quantified or vague? Did they bury a decline? Lenses target the Management Discussion & Analysis section; the IDs match our internal PR-MD-* taxonomy for release-note tracking.

PR-MD-1Choice
01

What drove the revenue change?

What mainly drove the revenue change this period?
volumepricemixacquisitionscurrencyone-time itemsnot disclosed
Why it matters. "Volume" vs "price" vs "one-time items" is the entire story on whether a top-line beat is durable. "Not disclosed" is itself a signal — vagueness is a red flag.
PR-MD-2Choice
02

What drove the margin change?

What mainly drove the margin change this period?
input costspricingmixscaleone-time itemsnot disclosed
Why it matters. Mix vs. one-time items changes how much of this quarter's margin to extrapolate. Scale-driven margins compound; one-time doesn't.
PR-MD-3BinaryYes = risk
03

Decline buried in MD&A?

Does MD&A disclose a decline in a key operating metric (customers, units, retention, backlog) that wasn't flagged in the headline?
Why it matters. Catches the "yes we grew revenue but net adds went negative" moment where the number investors actually care about is three paragraphs in.
PR-MD-4BinaryYes = signal
04

Specific, quantified outlook?

Does management give a specific, quantified outlook for the next period?
Why it matters. Guidance withdrawal (or sudden vagueness) is one of the earliest leading indicators of a bad quarter ahead. The binary makes it trivial to diff across releases.
PR-MD-5Rating 1–5
05

Margin-explanation specificity

How specific is the margin explanation? 1 = no drivers named · 5 = every driver quantified.
Why it matters. Disclosure quality trending. A 5 → 2 drop across quarters is a story on its own, even when the margin number itself didn't move much.
Software Contracts · Procurement & sales

Five lenses that read both sides of a software contract before legal does.

Different contract types, different sides of the table. The buyer wants to know about price caps, data-training rights, and lock-in. The vendor wants to know about scope-creep exposure. A pass of these five on a redline draft surfaces the dealbreakers in minutes instead of hours.

SaaS AgreementBinaryBuyer side
01

Uncapped renewal price?

Can the vendor raise the subscription price at renewal without a cap?
Why it matters. Year-2 cost shock is the most common SaaS buyer regret. Catches the missing "price increases capped at X%" language at redline time.
NDAChoiceEither side
02

Confidentiality duration

How long do the confidentiality obligations last after the agreement ends?
under 2 years2–5 yearsover 5 yearsindefinitelynot stated
Why it matters. "Indefinitely" and "not stated" are both red flags — one is a liability trap, the other is a legal void. Everything else is at least negotiable from a known baseline.
Professional Services / SOWRating 1–5Vendor side
03

Scope-creep exposure

How exposed is the vendor to unpaid scope creep? 1 = tightly defined deliverables with a change-order process · 5 = open-ended scope with no change mechanism.
Why it matters. A 4 or 5 means every out-of-scope ask gets delivered for free "to keep the relationship." The lens surfaces the risk before signing instead of month three.
Data Processing AddendumBinaryBuyer side
04

AI-training rights on customer data?

Is the vendor allowed to use customer data to train AI models?
Why it matters. Mirrors ToS #2 but on a B2B contract, where the opt-out is a redline not a settings toggle. Same question, higher stakes.
Supply / DistributionBinaryBuyer side
05

Exclusivity or minimum-purchase lock-in?

Does the agreement lock us into exclusivity or minimum purchase commitments we can't exit early?
Why it matters. The two clauses that turn a flexible supplier relationship into a strategic liability. Both are easy to miss unless you're specifically looking.
Shopify · SMB commerce

Five lenses a solo-admin merchant can wire into the order and support flow.

Shopify Analytics does dashboards; Shopify flags obvious fraud. Bernoulli's edge is calibrated reads on the messy, free-text parts a smart store manager would read if they had time: order notes, support messages, reviews, listing copy. Each lens below lives in a "Bernoulli says" card on the Order, Message, Review, or Product page — same shape as the Salesforce admin plugin, with Order and Customer standing in for Opportunity and Lead. Day-one install needs no Shopify approvals; a tier-two pack unlocks lifetime-value and churn lenses once protected-data scopes come through.

OrderBinaryRisk signal
01

Chargeback or dispute risk

Is this order likely to lead to a chargeback or dispute?
Why it matters. Chargebacks cost SMB merchants real money — the item is gone, the fee lands, the ratio climbs. A calibrated probability over order notes + address mismatch + buyer history flags the ~5% worth holding for manual review before fulfillment, without blocking the other 95%.
Support messageChoice
02

What does this customer want?

What is the customer actually asking for in this message?
refundreplacementinfocomplaint_onlycancel
Why it matters. Routes an inbox of mixed inbound into the four templates that handle 90% of it — refunds auto-queue, info gets an FAQ reply, complaint-only goes straight to the owner. Pairs with the urgency lens from the Salesforce pack if you want 2D triage.
ReviewBinaryRisk signal
03

Real defect, or one-off?

Does this review point to a product defect (consistent with other recent reviews), or a one-off?
Why it matters. One angry review is noise; three in a week on the same SKU is a signal. The lens reads the review alongside recent neighbors and flags SKUs that need a QC pass or supplier call before the return rate spikes.
Product listingRating 1–5
04

Listing clarity

How clear is this product listing — enough for a buyer to decide without asking a question? 1 = ambiguous or missing basics · 5 = complete and specific.
Why it matters. Low-clarity listings generate pre-sale support tickets and post-sale returns (sizing, color, materials). A rating sweep across the catalog surfaces the ten listings to rewrite first — the ones costing the most support hours per dollar of sales.
Discount codeBinaryRisk signal
05

Discount-code abuse

Does the recent usage pattern of this discount code look like abuse — shared addresses, cycling new-account signups, out-of-pattern velocity?
Why it matters. First-order discount codes leak on coupon sites and get drained by one or two actors cycling fresh accounts. The lens runs on the code's recent redemption pattern and flags codes to disable or tighten (e.g., require account age ≥ 30 days) before the budget evaporates.

Starter packs today. Your own tomorrow.

The five packs above ship with the app at app.bernoulli.live — they're the current examples, with more industries landing as we build. In the app, your team authors new lenses, groups them into packs, and points them at whichever source documents matter: a Chrome tab, a CRM record, an uploaded PDF, a Google Doc, an API feed. Calibrated probabilities, your questions, your data.