What Calibers.ai Is
Calibers.ai is a talent assessment platform built around a simple premise: hiring decisions should be based on evidence obtained following an academic peer reviewed framework, and not gut feeling. And the platform to execute that premise should be as simple and easy to use as possible, because as the number of candidates grow, assessments shouldn't be a bottle neck so recruiters have more time to meet the human behind the CV.
The platform is organized around four core functionalities, which together form a complete hiring evaluation pipeline:
◆ 1. CV Screening — Upload a candidate's CV and the platform evaluates it against your Caliber across seven structured domains: Knowledge, Experience, Education, Achievements, Career Trajectory, Character Signals, and Risk Signals. Each domain is scored from 0 to 100 and combined into a single CV Score using configurable weights. The result is an objective, structured reading of the candidate's background relative to what the role actually requires — available in seconds, consistently applied across every candidate.
◆ 2. Test Generation — Once a Caliber is defined, the AI generates role-specific assessments across up to four dimensions: functional knowledge, cognitive reasoning, personality (IPIP-HEXACO), and behavioral style (D/I/S/C). Tests are created from the Caliber's requirements — no question bank, no manual design, no external provider needed. A test that would take weeks to commission traditionally is ready in under a minute.
◆ 3. Test Execution — Each generated test is assigned to a candidate and delivered as a unique, secure link. No account is needed on the candidate's side. The test runs in a proctored environment monitored by camera, is AI-administered in a conversational format (for functional and cognitive tests), and cannot be paused or reused. The candidate's experience is professional and focused; the HR team's involvement during this phase is zero.
◆ 4. Reporting for Decision Making — The moment a candidate submits their assessment, the platform generates a full written report automatically. Reports cover scored breakdowns, strengths, weak areas, and an executive summary with a hiring recommendation — written in the language you choose, independent of the language the test was taken in. Results feed into a Match Score that combines CV and all completed assessments into a single Caliber-fit number for easy comparison across candidates.
Additionally, any Caliber can generate recruitment content — a hook-driven LinkedIn post and a full structured job description — with one click, for a small credit cost. See the Calibers section for details.
The Central Concept: The Caliber
Everything in the platform flows from one concept — the Caliber.
A Caliber is your definition of the ideal person for a specific role. It is not about any individual candidate; it is the aspirational profile you are hiring toward. You describe the job title, the responsibilities, the knowledge and skills you need, and the behavioral traits that matter. That description becomes the engine behind everything else: the AI reads your Caliber and uses it to generate tests, screen CVs, write job postings, and produce evaluation reports — all tailored to that specific role.
This means the most important work you will do on the platform is writing a good Caliber. A Caliber with clear, detailed responsibilities and a specific required knowledge section produces sharper tests, more accurate CV scores, and more relevant reports. A vague Caliber produces generic output. The AI is only as specific as the input you give it.
Once a Caliber exists, deploying it is fast. Generating a test takes seconds. Assigning it to a candidate takes a few clicks. A candidate can receive an assessment link, complete the test, and have a full report in your dashboard — all within the same day.
How the Dashboard Is Organized
The HR dashboard is a single page with a horizontal tab strip — not a sidebar. When you first enter, you land on a home screen with four large numbered cards laid out as a workflow:
- Calibers — Define the skill set and knowledge profile you want to evaluate.
- Candidates — Add and screen the people you are evaluating.
- Tests — Generate role-specific assessments.
- Pools — Group candidates to compare them against a Caliber.
Clicking a card jumps you to that section. Once inside, the same four areas appear as a tab strip at the top — Calibers · Candidates · Tests · Pools — and you can switch freely between them. A Start button (to the left of the tabs) returns you to the home screen at any time.
The top navigation bar (above the dashboard) is separate: it holds Docs, Help, account, share/referral, and logout — not the four work areas.
The Five Core Workflows
These are the five tasks you will perform most often. Each is a verified, exact click-path. The rest of the guide expands on each area in depth.
1. Create a Caliber
- Open the Calibers tab.
- On the My Calibers subtab, click + New Caliber (top right).
- In the New Caliber modal, on the Profile tab, fill in:
- Caliber Name
- Responsibilities
- Required Knowledge
- Click Save Caliber.
That's the minimum. You can refine screening and scoring weights afterward on the modal's other tabs (see Calibers).
2. Screen or Ingest Candidates (CV)
- Open the Candidates tab.
- In the toolbar, use the Parse / Screen control:
- Parse (default) — extract the candidate's details from the CV, no scoring.
- Screen — click it to also score the CV against a Caliber.
- Drop CV files onto "Drag your candidates' CVs here, or click to add manually" (PDF, Word, or image; multiple files allowed).
- A configuration modal opens:
- Screen mode → Screening Configuration: pick the Caliber to screen against, the Report language, and how to treat duplicates (Skip or Update).
- Parse mode → a confirmation that candidates will be added without a CV alignment score.
- Click Continue / Continue Parsing. A credits confirmation appears showing the exact cost — click Yes, continue.
- When it finishes, the Upload Complete summary lists each candidate with a status (Added / Updated / Skipped / Error) and, if screened, a CV Score.
You can also create a candidate manually with + New Candidate and type their details in.
3. Generate a Test
- Open the Tests tab.
- Stay on the My Tests subtab (the default).
- Click + New Test (top right).
- In the Generate Test with AI modal, choose:
- Test Type — Technical, Cognitive, Personality, or Behavioral.
- Caliber — the role this test is for (required for Technical and Cognitive; not used for Personality or Behavioral).
- Test Language — the language the candidate takes the test in.
- Plus type-specific options (seniority, number of questions, focus/duration — see Tests).
- Click Generate. Generation takes a few seconds up to a minute. On success you'll see "Created: <title>" and the new test appears in the My Tests list, ready to assign to as many candidates as you want.
4. Assign a Test to a Candidate
- Open the Candidates tab and click a candidate to open their panel.
- Go to the candidate's Tests sub-tab.
- In the Assign Test section, choose a test from Assign a test:. (The Test Language field is read-only — it mirrors the test's own language. Report Language is yours to set and is fully independent of the test language.)
- Click Assign Test.
- An Email this assessment window opens. Either:
- Send email — the platform emails the secure link to the candidate, or
- I'll share it myself — close the window and copy the link yourself.
- To copy the link manually later: in the candidate's Tests sub-tab, each assigned session has a Link button (it shows Copied! after you click).
You can also assign at creation time (the + New Candidate modal has an Assign a test dropdown), or from the Tests tab: on a test row click Assign, then Assign Test, and search the candidate by name or Candidate ID. If no candidate matches, you can add them (first and last name) right there — no need to leave the window.
5. Organize Candidates in Pools
- Open the Pools tab and click + New Pool.
- In the New Pool modal, enter a Pool Name, optionally associate a Caliber, and pick a Color. Click Create Pool.
- Open the pool and use the Members tab to add candidates two ways:
- Search an existing candidate and click Add to Pool, or
- Drop a CV on the pool's own drop zone — the candidate is created and added to the pool automatically (and screened against the pool's Caliber if Screen mode is on).
- You can also toggle a candidate into any pool from the candidate's own panel, under its Pools tab (a checklist of all pools).
When a pool has an associated Caliber, its Members table shows each candidate's Match Score, sortable, so you can rank the group against the role.
Calibers
What a Caliber is: a job-role profile — your definition of the ideal candidate for a position. It drives test generation, CV screening, job-post generation, and reporting. One Caliber describes one role; many candidates are evaluated against the same Caliber.
Where it lives: the Calibers tab, which has two subtabs — My Calibers (your roles) and Calibers Library (starter templates you can copy).
Creating and editing: + New Caliber opens the New Caliber modal; clicking an existing row opens it as Edit Caliber. Save is Save Caliber when new, Save Changes when editing. A Delete button appears when editing.
The four tabs inside a Caliber:
| Tab | What it's for |
|---|---|
| Profile | The core definition: Caliber Name, Responsibilities, Required Knowledge. These three drive everything the AI produces for the role. |
| Screening Settings | The seven CV domain weights — how much each screening domain counts toward the CV Score (see Weights). Includes an AI Auto Tuning button that suggests weight levels from your Caliber's content (costs credits). |
| Evaluation Settings | The assessment dimension weights — how much CV, Technical, Cognitive, Personality, and Behavioral each count toward the Match Score (see Weights). |
| Candidates Alignment | (Editing only.) A scored table of every candidate evaluated against this Caliber — Match, CV, and per-test scores side by side. |
Screening domains: when you screen a CV against a Caliber, the AI scores the candidate across seven domains, each 0–100:
| Domain | What it measures |
|---|---|
| Knowledge | Subject-matter expertise and technical know-how matching the role |
| Experience | Depth and relevance of hands-on work history in similar positions |
| Education | Formal academic background, degrees, certifications, training |
| Achievements | Concrete results, impact metrics, notable accomplishments |
| Career Trajectory | Whether the career path shows progressive growth toward this type of role |
| Character Signals | Soft skills, leadership indicators, work ethic and values cues in the CV |
| Risk Signals | Concerns — gaps, very frequent job changes, inconsistencies. Scored inversely: 100 = no risk detected; lower = more concerns |
Generate recruitment content: on the Profile tab, the Generate Social Post button (costs credits) reads the Caliber's title, responsibilities, and required knowledge and produces two pieces of content in the Caliber's own language:
- A short, hook-driven social post (LinkedIn-style) that opens with an attention-grabbing statement rather than a standard "we're hiring" line.
- A complete, structured job description with sections for Role Summary, Responsibilities, Requirements, Location & Work Model, and How to Apply. Placeholders (e.g.
{{Location}},{{email}}) mark anything not present in the Caliber.
You can copy or edit either piece before use.
Calibers Library: the Calibers Library subtab (next to My Calibers) holds a curated set of ready-made example Calibers, published by Calibers.ai, that you can browse and Copy to My Calibers to use or adapt as a starting point. New accounts are pre-seeded with sample Calibers — and sample Tests — in the account's language at registration. The Tests tab has its own Tests Library subtab that works the same way for ready-made tests.
Candidates
What a candidate is: a person being evaluated. Each candidate is identified by a unique Candidate ID (for example C-20260720-001), assigned automatically when the candidate is created — this is the platform's identifier for the candidate. They also have a name, an optional email, an optional CV, and scores accumulated from CV screening and tests. Email is only needed when you want the platform to email the exam link on your behalf; otherwise you can copy and share the link yourself.
Where they live: the Candidates tab. Clicking a candidate opens a panel with sub-tabs: Calibers (per-Caliber scores and match breakdown — the default tab), Analyzer (AI scoring against a Caliber — CV, Personality, and Behavioral), Personal Information, Pools (toggle pool membership), and Tests (assign and track sessions).
Creating a candidate:
- Manually — + New Candidate opens the New Candidate modal (name, last name, optional email, profession, phone, CV language, and more). Only first and last name are required. You can also assign a test right there via the Assign a test dropdown, and add the candidate to pools with the Pools checklist. Submit with Add Candidate.
- From a CV — drop one or more files on the Candidates drop zone; the AI extracts name, contact details, and experience. See Workflow 2.
Parse vs Screen:
- Parse — extracts structured details from the CV. Costs credits, and an image or scanned PDF costs more because it has to be read with OCR first.
- Screen — additionally scores the CV against a chosen Caliber across the seven domains and writes a report, for an added cost on top of the parse. You choose the Caliber and report language in the Screening Configuration modal.
File handling & privacy: original CV files are not stored. The platform extracts the text and processes it with AI — up to 7,000 characters per file (more than double the length of an average CV); extremely long CVs may be truncated. Accepted formats are PDF, Word, and image files.
Duplicate handling: when a CV's email already matches an existing candidate, you choose Skip (leave the existing record) or Update (overwrite it with the new data). In batch uploads this is set once in the config modal; for a single conflicting CV you'll be prompted with Keep editing vs Update existing candidate. There is no separate "merge" or "create a second copy" option — the two choices are skip or update. Duplicate matching relies on the email address, so candidates added without an email can't be auto-deduplicated — each such CV creates a new candidate record.
Where CV Scores appear: in the Candidates table's Calibers column, as a badge in the form Caliber Name · 82%, colored by score. A candidate screened against several Calibers shows one badge per Caliber, since each score is relative to that Caliber.
Re-screening: you can re-screen any existing candidate against a Caliber at any time from the candidate panel's Analyzer tab → Scoring section (the CV Scoring button). A re-screen never re-runs OCR, so it costs less than screening a freshly uploaded scanned CV.
Tests
What tests are: assessments delivered to a candidate through a secure link. Results feed the candidate's Match Score. A generated test is saved as a reusable item under the Tests tab → My Tests subtab; you can assign the same test to as many candidates as you like.
Where tests live: the top-level Tests tab has three subtabs — My Tests (tests you've generated), Tests Assigned (sessions handed to candidates), and Tests Library.
Generating a test
Click + New Test on My Tests to open the Generate Test with AI modal. The fields shown depend on the Test Type:
| Field | When it appears | Options |
|---|---|---|
| Test Type | Always | Technical · Cognitive · Personality · Behavioral |
| Test Focus | Technical only | Operational Knowledge · Developer Focused |
| Caliber | Technical & Cognitive (required); disabled for Personality/Behavioral | Your existing Calibers |
| Seniority | Technical only (disabled for Cognitive/Personality/Behavioral) | Junior · Mid · Senior |
| Number of Questions | Hidden for Behavioral | Technical: 10/15/20/30 · Cognitive: 20/30/40 · Personality: 60 or 96 |
| Test Duration | Developer-Focused Technical only | 30–240 minutes |
| Test Language | Always | The language the candidate takes it in |
Click Generate. On success you see "Created: <title>" and the test lands in My Tests. Each test row offers Assign (assign candidates or see who's assigned), YAML (inspect), Duplicate, and Delete (only when it has no active sessions).
The four test types
- Technical (also called Functional) — a skill-based test tailored to the role. Technical and Functional name the same test type; which word fits is just a matter of the role. Technical is the natural term for operational, hands-on roles — engineering, software development, IT. Functional better describes knowledge-based roles in areas like finance, accounting, or administration. In the platform the test type is labeled Technical when you generate it, and its report scores this dimension as Technical / Functional Knowledge. Questions are generated from the Caliber's responsibilities and required knowledge. It is AI-administered in a conversational format: a virtual examiner presents questions one at a time and the candidate types answers (a mix of open-ended and multiple choice). Seniority adjusts depth: a senior test probes architectural thinking and edge cases; a junior test focuses on fundamentals. For Technical tests you also choose a Test Focus: Operational Knowledge tests the tools, frameworks, and technologies named in the Caliber via an AI-conducted conversation with mixed question types; Developer Focused delivers language-agnostic algorithmic challenges balanced by difficulty and timed to a Test Duration you set.
- Cognitive — a domain-independent reasoning assessment. It measures how a person thinks, not what they know — logical deduction, sequences, analogies, spatial reasoning, pattern transformations. All multiple choice, single correct answer. Its structure is the same regardless of Caliber, which makes it a clean, unbiased measure of aptitude. Use it when raw reasoning matters — analytical roles, fast-learning environments, novel-problem work.
- Personality (IPIP-HEXACO) — a validated psychometric survey measuring stable personality traits across six dimensions, scored relative to published population norms. Available in 60 questions (~20 min) or 96 questions (~30 min). IPIP-HEXACO tells you who the person fundamentally is.
| Factor | Letter | What it captures |
|---|---|---|
| Honesty-Humility | H | Sincerity, fairness, modesty, resistance to self-serving behavior |
| Emotionality | E | Emotional sensitivity, empathy, need for support, stress response |
| eXtraversion | X | Social confidence, energy, enthusiasm |
| Agreeableness | A | Forgiveness, patience, flexibility, conflict management |
| Conscientiousness | C | Organization, diligence, thoroughness, deliberate decisions |
| Openness to Experience | O | Curiosity, creativity, comfort with unconventional ideas |
60 vs 96 questions: both versions measure the same six factors. The 96-question version simply measures each one with more items, which makes the scores more reliable. Choose 96 when precision matters most, and 60 when you want to keep the time commitment shorter. How scores are read: each factor is scored on a 1–5 scale and then compared against published population norms (Ashton, Lee & Goldberg, 2007), so the report tells you not just the raw score but whether it sits above, below, or near the general-population average.
- Behavioral (D/I/S/C) — a forced-choice assessment of how a person prefers to operate at work — their behavioral style, as opposed to IPIP-HEXACO's stable traits. Across 20 blocks, the candidate sees four statements (one per style) and picks which is most and which is least like them — a format that reveals natural preferences rather than a socially desirable self-image.
| Style | Letter | What it captures |
|---|---|---|
| Dominance | D | Assertiveness, directness, results-driven drive |
| Influence | I | Sociability, enthusiasm, persuasion, people-orientation |
| Steadiness | S | Patience, reliability, collaboration, consistent pace |
| Conscientiousness | C | Precision, analytical approach, adherence to quality standards |
The report then classifies each of the four styles by how strongly the candidate selected or avoided it across the blocks (net Most-minus-Least):
| Classification | What it means |
|---|---|
| Primary | Highest net score — the candidate's dominant behavioral tendency |
| Secondary | Second-highest — a strong supporting style |
| Supporting | Third-highest — present but not dominant |
| Suppressed | Lowest — the style the candidate most moves away from |
DISC results are relative within the candidate — they rank the four styles against each other, not against a population. No profile is universally "good" or "bad."
IPIP-HEXACO vs D/I/S/C — which to use? IPIP-HEXACO measures stable traits against population norms and is best for predicting long-term fit, integrity, and performance. DISC measures behavioral style against the candidate's own other styles and is best for understanding communication and day-to-day work approach. For the fullest picture, run both.
One test = one language. A test is generated in a single language. To offer the same assessment in another language, generate a new test in that language.
Assigning a test and the exam experience
See Workflow 4 above for the exact steps. Key points:
- Report Language is chosen at assignment, not at generation, and is independent of the test language. To change a report's language, unassign and reassign with the correct Report Language.
- Assigning reserves the test's finishing cost from your balance, to guarantee the candidate can finish and be scored (see Credits → Credit reservation).
- The candidate opens a secure, proctored environment via the link — no login needed. The camera monitors the session; the test cannot be paused, resumed, or reused. The link expires once used or after the session times out.
- Managing a session (from the candidate's Tests sub-tab, or the Tests Assigned subtab): Link (copy), Report (view once available), Unassign / Delete (removing a session permanently deletes any generated report — you'll be asked to confirm).
Reports
When a candidate submits, the platform automatically generates a written report in the Report Language you chose (independent of the test language). Reports are viewable in the dashboard and downloadable as PDF. Every report closes with an Executive Summary and an overall verdict — Aligned / Partially Aligned / Not Aligned.
What each report contains:
- Technical — a per-question Score Summary table (topic, type, score, rationale) with a total; then narrative sections on Technical / Functional Knowledge, Communication (how clearly the candidate expressed their thinking on open-ended answers), Problem-Solving, Strengths, and Weak Areas, each referencing actual answers; then the Executive Summary and verdict. (These are hiring assessments, not coaching reports — weak areas are named, not prescribed fixes.)
- Cognitive — the same Score Summary table, then Reasoning Category Breakdown, Problem-Solving Approach, Strengths, Weak Areas, and the Executive Summary. Focused on reasoning quality rather than domain knowledge.
- Personality (IPIP-HEXACO) — a Factor Score Summary, then one section per factor, each with two parts: a pure interpretation of the score in a workplace context, followed by a comparison to the general population (above / below / near average). A Cross-Factor Patterns section and the Executive Summary close it, calling out factors that sit notably outside the population average.
- Behavioral (D/I/S/C) — a factor table (Most / Least / Net counts and the Primary/Secondary/Supporting/Suppressed label per style), a per-style interpretation of what the pattern means for the candidate's work approach, Strengths, Weak Areas, and the Executive Summary.
Pools
What a pool is: a collection of candidates grouped for one evaluation context. A pool can be tied to a Caliber; that association is optional but is what lets the pool rank members by Match Score.
Where they live: the Pools tab — a table with Name / Caliber / Members / Created, plus a search box and a Caliber filter. Clicking a pool opens its panel on the Members tab.
Creating a pool: + New Pool → New Pool modal → enter Pool Name, optionally select a Caliber, pick a Color, then Create Pool.
Adding candidates (three ways):
- In the pool's Members tab, search an existing candidate and click Add to Pool.
- In the pool's Members tab, drop a CV on its drop zone — the candidate is created and added, and screened against the pool's Caliber if Screen mode is on.
- From a candidate's own panel, under its Pools tab, tick the pool in the checklist. A candidate can belong to many pools at once.
Ranking: when the pool has an associated Caliber, the Members table shows a sortable Match Score column, so you can rank the whole group against the role. In the Candidates table, pool membership shows as colored pool chips in the Pools column.
Why the same candidate can score differently across pools: their CV Score is always relative to a Caliber, so a candidate in two pools tied to two different Calibers can hold two different scores.
Scores
Two distinct scores exist for a candidate. They serve different purposes and are calculated differently — never conflate them.
CV Score
The output of screening a candidate's CV against a Caliber. Always 0–100.
- How it's calculated: the AI scores the seven domains (Knowledge, Experience, Education, Achievements, Career Trajectory, Character Signals, Risk Signals), and the final score is the weighted average using the domain weights on the Caliber's Screening Settings tab:
CV Score = Σ(domain_score × domain_weight) / Σ(domain_weights)
- When: once, at screening time. It's stored and doesn't change unless you re-screen (which costs credits).
- Where: pool/Caliber badges, e.g. "Backend Engineer · 78%".
Match Score
The overall measure of how well a candidate fits the role end-to-end. It combines every assessment that has a result — CV screening, Technical, Cognitive, Personality, Behavioral — into a single 0–100 number.
- How it's calculated: a weighted average of the available results, using the dimension weights on the Caliber's Evaluation Settings tab:
Match Score = Σ(score × weight) / Σ(weights of completed dimensions only)
- Only scored dimensions count. If a candidate has just a CV Score and a Technical result, only those two are averaged — the others are excluded from the denominator until they exist. No assessment is mandatory.
- How each dimension's score becomes available — this is not the same for all four tests. Technical and Cognitive are scored automatically the moment the candidate submits the test, and flow into the Match Score with no further action. Personality and Behavioral are different: submitting the test produces the report, but not a Caliber-alignment score, so completing the test alone adds nothing to the Match Score. To make either one count, you run its alignment scoring from the candidate's Analyzer → Scoring section — the Personality Scoring and Behavioral Scoring buttons. Each asks you to pick a Caliber, costs credits, and scores that profile against the specific role. Until you run it, that dimension stays out of the Match Score even though the test is complete. (The reason: a raw IPIP-HEXACO or D/I/S/C profile is not "good" or "bad" on its own — it only becomes an alignment number relative to a role.)
- When: live, recomputed every time you view the candidate — never stored. Early on (CV only), Match Score equals CV Score; it evolves as more dimensions are scored.
| CV Score | Match Score | |
|---|---|---|
| Measures | CV only — background on paper | CV + all completed tests combined |
| Updates | Once, at screening | Live, on every view |
| Caliber scope | Tied to one Caliber | Tied to one Caliber |
| Missing assessments | N/A | Excluded from the average — no penalty |
A worked example
Suppose a Caliber weights its dimensions CV 1.0, Technical 0.8, Personality 0.4, and a candidate has a CV Score of 80 and a Technical score of 60. The candidate may have already completed the personality test, but until you run its alignment scoring in the Analyzer there is no personality score to include. Only scored dimensions count:
Match Score = (80×1.0 + 60×0.8) / (1.0 + 0.8) = 128 / 1.8 ≈ 71
Once you run the personality alignment in the Analyzer (say it scores 90), it joins the average:
(80×1.0 + 60×0.8 + 90×0.4) / (1.0 + 0.8 + 0.4) = 164 / 2.2 ≈ 75
The CV Score itself is computed the same way — a weighted average, but across the seven screening domains instead of the assessment dimensions.
Weights
A weight controls how much a piece of information influences a score. Higher weight = more influence. The platform uses weights in two places, both set inside the Caliber:
- CV domain weights — on the Caliber's Screening Settings tab. They tell the platform which of the seven CV domains matters most for this role (e.g. weight Education high for a junior role, Achievements high for a senior one). Set with a per-domain slider from level 1 (least important) to level 5 (most important) — level 1 still counts, it is never fully ignored (each level maps to a multiplier from 0.2 at level 1 to 1.0 at level 5). The AI Auto Tuning button (costs credits) suggests a level for each domain based on the Caliber's content; accept, adjust, or override it.
- Assessment dimension weights — on the Caliber's Evaluation Settings tab. They decide how much each assessment type (CV, Technical, Cognitive, Personality, Behavioral) contributes to the Match Score. Raise a dimension's weight to make that result pull the overall score more strongly.
Credits
What credits are: the platform's unit of consumption. Every AI operation costs credits. You spend them; you don't earn them back from normal use (except via the referral program).
Where credits come from:
- Subscription plan — a monthly allowance that resets each billing cycle; unused subscription credits don't carry over.
- Credit packs — one-time purchases that stack on top of your subscription and expire after 90 days.
- Privileged Account (Head Start Super Pack) — a never-expiring credit pack.
Spending order: subscription credits first, then the pack expiring soonest.
What costs credits — every AI operation does. The confirmation dialog always shows the exact cost before you commit, so that dialog, not this guide, is the number to trust:
- CVs — parsing a CV is the cheapest operation. Screening it against a Caliber adds to that, and a scanned or image CV adds an OCR charge on top because it must be read before it can be parsed.
- Tests — generating a test costs more than a CV operation; cognitive tests are the most expensive to generate. Finishing a test (writing the report) is charged separately from generating it.
- Smaller helpers — generating a social post, AI Auto Tuning for CV weights, and pool analyses are all at the low end.
Credit reservation: when you assign a test, that test's finishing cost is immediately reserved (earmarked) to guarantee that candidate can always complete and be scored. Reserved credits are held aside — your Available balance = physical balance − reserved. Other operations can only spend Available credits. The credits menu breaks out Balance / Reserved / Available.
When credits run low: if your Available balance drops below the platform minimum, your account enters a grace period to buy more or subscribe. After the grace period, AI features are suspended until you recharge.
Settings & Your Account
Open the account modal from the account control in the top navigation bar. It has four tabs:
- Profile — change your first name, last name, email, and UI language (English, Spanish, Portuguese, Japanese, Chinese). Your UI language sets the console's language and the language of the verification and invite emails your organization sends.
- Security — change your password.
- Billing — your billing address and tax ID (org admins only). See Billing below.
- Team — invite and manage members, and set the org's default language. See Team Management below.
Marketing Preferences (org admins) — a separate control in the page footer (the "Marketing Preferences" link), not in the account modal. It toggles whether Calibers.ai may use your data for marketing analytics; turning it on is what allows a Google Ads conversion tag to fire. It does not affect any platform functionality.
Member vs. admin: org members can change their own name, email, password, and UI language only. Org admins can additionally edit billing details, the org's default language, and team membership.
Billing
- Payments are processed by Polar (our Merchant of Record, built on Stripe). We never see your payment method; Polar handles invoicing and tax.
- Buying credits: open the credits menu in the top bar → choose a credit pack or subscription → complete checkout on Polar → credits apply automatically (or click Sync).
- Subscriptions renew monthly and can be canceled anytime; access continues until the end of the paid period.
- Billing details (org admins): Account → Billing Information — name, address, country, tax ID. These pre-fill your next checkout.
- Invoices are sent by Polar to your billing email after each payment.
Team Management
Calibers.ai supports multiple users sharing one account and one credit pool.
- The person who registers the organization is the Org Admin; they can invite colleagues from Account → Team by entering an email and clicking Invite. The invitee gets a link that creates their account with no separate verification step, joining as an Org Member.
- Org Admin: full access — billing, team, org language, and all HR features.
- Org Member: all HR features (create candidates, run tests, screen CVs) but cannot change billing, invite others, or remove members.
- Shared credit pool: every member's AI operations draw from the same balance.
- Admins can promote a member to admin or remove them from the Team tab.
Referral Program
Share your personal referral link (the Share control in the top bar). When someone registers through it and makes their first purchase, you receive 50 credits and they get a 7% discount on that purchase. Your referral code (e.g. XM83J) is embedded in the link and captured automatically at registration.
Support
Every account has Airelia-1, the Calibers.ai AI support agent, available 24/7 from the chat widget in the bottom-right corner. Ask it anything about how the platform works — features, scores, billing, tests, candidates, credits — and it replies in the language you write in (English, Spanish, Portuguese, Japanese, Chinese).
If Airelia-1 can't resolve something — a technical problem, an account matter, anything needing a human — it offers to open a support ticket that automatically includes your full conversation, so the team has the context without you re-explaining. You'll get a confirmation email with a case number, and a team member follows up by email.
Common Issues
I didn't receive my verification email — check spam/junk for mail from hello@airelia.ai. If it's not there after 10 minutes, register again with the same email; the platform resends a fresh link.
A candidate didn't receive their exam link — verify the email on the candidate's record and ask them to check spam (sent from hello@airelia.ai). You can also copy the link manually from the candidate's Tests sub-tab (the Link button) and share it directly.
My credits didn't update after purchase — click Sync in the credits menu; if needed, wait a minute and sync again.
I have credits but the system says I don't have enough — you likely have Reserved credits tied to pending exams. The credits breakdown shows Balance, Reserved, and Available separately; only Available can be spent on new operations.
An exam link expired — go to the candidate's Tests sub-tab, unassign the current session, and assign a new one to generate a fresh link.
The report language is wrong — Report Language is chosen at assignment, not at generation. Unassign and reassign the test, selecting the correct Report Language.