Voice.
The Evalgist voice rules in readable form. Every rule on this page is rendered from @evalgist/content-style. The full guide for AI tools is at the end of the page and at /brand/voice.md.
Substantiation, not conclusions.
The ceiling for every surface. Below it: the job the product does, the readers it serves, and how it differs from a chat window.
Signals: One criteria set · Quoted evidence · Exportable record
Job to be done
JOB_TO_BE_DONEAudience
AUDIENCEThe contrast
VS_AD_HOC_CHAT- A chat can score one resume. It will not hold one standard across two hundred, or hand you a packet for the meeting.
- A chat window starts every conversation from zero. A criteria set holds the same standard for every document in the batch.
- You can ask a chat why it ranked someone. You cannot hand its answer to the committee.
- Better than ChatGPT.
- Smarter than a chatbot.
- Don't trust generic AI.
Sober, confident, slightly dry.
Short sentences. English throughout, with US spelling: analyze, summarize, labeled. Confidence comes from the evidence, not from the adjectives.
Register
TONERhythm
cadencePunctuation
PRODUCT_UI_COPY.punctuation- Separate facts in a status line with a middle dot: '12 resumes ready · 2 not added'.
- End in-progress labels with an ellipsis character: 'Starting analysis…'.
- Use the Oxford comma: 'PDF, DOCX, or XLSX'.
- Refer to UI labels capitalized and without quotes: 'Use Revert edits to restore the cleaned version.'
Name the work, not the excitement.
Prefer verbs for what the reviewer does and nouns for what the workflow produces. The avoided list covers hype, filler, and claims the product cannot hold.
- evidence
- quoted evidence
- one criteria set
- batch
- evidence trail
- audit trail
- exportable record
- evaluate
- shortlist
- grade
- consistent
- consistent standard
- faster
- per criterion
- per candidate
- per answer
- review
- decide
- verify
- professional
- human decision
- revolutionary
- game-changing
- unleash
- supercharge
- seamless
- seamlessly
- easily
- effortlessly
- 10x
- cutting-edge
- next-generation
- disruptive
- bias-free
- objective AI
- fully compliant
- AI sparkle
- magical
- CV
- leverage
- powerful
- intelligent
- Let's
- dive into
Product stoplist checks import AVOIDED_WORDS_PATTERN: whole words, case-insensitive, plural s included, plus emoji. Do not keep a local list.
AI language
AI_POSITIONING · SEO_AI_LANGUAGE- AI-assisted evaluation
- AI resume screening with quoted evidence. Every match cites the resume line it rests on.
- AI resume screening with quoted evidence.
- AI-assisted evaluation that keeps the human as the decider.
- AI document review that cites the source line behind each judgment.
- The AI extracts, compares, and cites evidence. The reviewer decides.
- Powered by AI
- AI-powered hiring that transforms screening.
- Powered by AI.
- Objective AI candidate screening.
- Fully compliant AI evaluation.
- Automated hiring decisions.
Search terms: AI resume screening · AI-assisted resume screening · AI candidate screening · AI-assisted evaluation · AI document review · AI-assisted document review · AI grading · AI-assisted grading
Full name first. Suffix after.
Products are never sub-brands. Outside the product, the first mention uses the full name. Inside the product, the sidebar lockup counts as the first mention.
Products
PRODUCT_NAMINGAuth surfaces
PRODUCT_NAMING.authSurfacesLabels describe evidence, not verdicts.
Shortlist holds the reference labels. Other products keep the same states with their own nouns.
Criterion status
PRODUCT_UI_COPY.criterionStatusStatus labels describe the evidence, not a verdict. Never write 'Met' or 'Not met' in labels, help text, or summaries. Exception: an audit surface that explains the score calculation may use the scorer's words (met, partial, not met) so its counts can be checked row by row.
Bands and decisions
matchBands · decisions · verdictErrors and toasts
errors · toastsBilling
PRODUCT_UI_COPY.billing- In Shortlist, one credit analyzes one resume.
- Say 'free credits' for the starter grant, not 'free analyses'.
- Say a failed analysis is 'refunded': the credit returns to the balance.
- The action label for buying is 'Buy credits'.
Claim only what the workflow produces.
Every trust claim is public. Write it for a customer, a candidate, an auditor, or a regulator. Reason: GDPR + EU AI Act audit-caution.
Retention and defensible language
DEFENSIBLE_CLAIM_RULESClaim boundaries
COPY_SURFACE_RULES.claimBoundariesEvery surface
COPY_SURFACE_RULES- Anchor evaluation language in evidence, criteria, and human decision-making.
- Label AI suggestions and human decisions as different states.
- Name AI as the method only when it explains a visible function.
- Use SEO-visible AI terms when they match real search intent, then bind them immediately to criteria, quoted evidence, records, or human review.
- Use the full product name on first mention outside the authenticated product frame. Auth surfaces are the exception: they use the parent name Evalgist only.
- Treat every trust claim as a public claim that may be read by a customer, candidate, auditor, or regulator.
- State at least one concrete privacy or data-handling fact on every external surface: retention, residency, deletion, or what leaves the platform. Never a bare 'we take your privacy seriously'.
- Promise bias elimination, objectivity, complete compliance, or automated correctness.
- Use startup acceleration language to make evaluation work sound effortless.
- Make the customer, candidate, regulator, or protected group the butt of the joke.
- Replace a concrete source, quote, criterion, or workflow detail with generic AI capability language.
Multi-pass tiers
MULTI_PASS_CLAIM_RULES- Never claim independence between passes. Passes run on related model families and share blind spots; describe them as cross-checked, never as independent reviewers or as models that 'agree'.
- Never imply an extra pass improves the score, grade, or ranking. Sell what it produces: the criteria worth the reviewer's time.
- Name the tier after the process, not the outcome. 'Cross-checked' describes what ran; 'Verified' asserts a result the product cannot stand behind.
- Never name a model in public copy. A model is a config value that changes; the processor behind it is named in the subprocessor register, which is where that question belongs.
Trust beats
TRUST_FINEPRINTThe full guide, for AI tools.
Paste this into an AI tool before drafting Evalgist copy. It holds every rule above, plus landing formulas, long-form registers, and the review checklist. It is generated from the same package.
Raw markdown: /brand/voice.md