Launching soon. CAI-1 “AI Foundations” will be free, for everyone.Launching soon · CAI-1 is freeSee the credential ladder →
Certavant

AI skills, demonstrated

CAI-1 AI Foundations: assessment blueprint

Assessment blueprint · Version 0.2.0-draft

Review and limits. The course content has completed cross-model editorial review. The task descriptions and operational pass rule below describe the configured final exam. Domain weights, timings and the pass standard remain provisional: no completed job-task analysis or empirical standard-setting study is claimed.

What CAI-1 shows

A CAI-1 credential shows that the holder briefed, checked and refined AI work in a recorded task and completed the other tasks with permitted tools, using everyday work that applies to any role. The course teaches how to brief an assistant, judge and verify what it produces, handle information responsibly, and recognize when AI is the wrong tool.

CAI-1 is performance-based. You earn it by doing realistic work tasks with AI, not by recalling facts about AI. It is genuinely failable: completing the lessons does not earn it.

CAI-1 is the same for every role. Role-specific skills are certified at CAI-2.

Competency domains and weights

| Domain | Weight | A competent candidate can… | | ----------------------------------------- | ------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | D1. Delegating work to AI | 30% | write briefs with a clear goal, audience, inputs, constraints and definition of done; break larger tasks into checkable steps with human decisions kept human; refine output with specific feedback; recognize when to start a fresh conversation. | | D2. Evaluating and verifying output | 48.33% | recognize common failure patterns (invented specifics, calculation errors, misread sources, omissions, additions beyond the brief, agreement with the user's framing); match checking depth to the stakes; verify against independent sources; check summaries against their source; keep a verification record. | | D3. Data, privacy and responsible use | 16.67% | classify information before sharing it; choose tools that are approved for that information; minimize and de-identify data; recognize untrusted content that may carry instructions; keep decisions about people under human judgment; own and, where appropriate, disclose AI-assisted work. | | D4. Choosing tools and knowing limits | 5% | explain at a practical level what an AI assistant does and doesn't know; distinguish tasks inside and outside the "jagged frontier"; choose a suitable kind of tool for a task; decide when not to use AI. |

The three final tasks are weighted equally. Each criterion is counted once in its primary domain. The domain weights are the mean of these task-level shares, rounded for display:

| Final task | D1 | D2 | D3 | D4 | | ----------------------------- | --: | ---: | --: | --: | | Brief the workspace assistant | 75% | 25% | 0% | 0% | | Check the sales summary | 0% | 100% | 0% | 0% | | Plan the survey analysis | 15% | 20% | 50% | 15% |

Task types

Course practice uses four task types. Its feedback helps you prepare for the three final tasks described below; it does not decide a credential.

| Task type | What you do | How it's scored | Main domains | | ------------------------- | -------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------- | ------------ | | AI practice (sandbox) | Complete a real task with an AI assistant: brief, review, refine, deliver. | Formative feedback on brief quality, refinement, accuracy and ownership notes | D1, D2 | | Critique | Find the errors, unsupported claims and risks in an AI output, and justify each finding against the source material. | Formative feedback on findings, source-grounded corrections and avoiding false alarms | D2, D3 | | Scenario | Make a sequence of decisions in a realistic workflow, with consequences. | Deterministic, from the quality of each choice | D1–D4 | | Work product | Produce a short artifact, such as a plan, a verification log or a data-handling plan. | Formative feedback against analytic criteria and their behavioral descriptions | D1–D4 |

How you earn CAI-1

You earn the credential by passing the CAI-1 final assessment. Course attendance or completion alone does not earn it. The course includes five module checkpoints at the mastery threshold (80%), with unlimited retakes and feedback after every attempt; these are practice. You may take the final without completing the course first.

The final assessment

  • Format: three short performance tasks in the browser-based assessment workspace, drawn from the course's task families. One is an AI-practice task, one a critique task, and one a work-product task involving a data-handling decision.
  • Time: one shared 60-minute window for all three tasks. There is no separate timer for each task; you choose how to divide the time. Timing remains provisional pending study, but 60 minutes is the configured limit.
  • Task contexts: the initial CAI-1 final uses one fixed, reviewed context for each of its three tasks. Candidates may receive the same company, figures and constraints. Sharing task content is prohibited. Published course practice instances are not final-assessment instances.
  • What's graded: your work products, including requested verification and data-handling notes. Task 1 also grades your recorded conversation with the provided workspace assistant: how you brief, check and refine its output. Complete that task in the workspace assistant; an external conversation cannot replace the recorded evidence.
  • Integrity: honor-code attestation at the start of the assessment.

The initial final uses these three reviewed task contexts (one fixed context per task):

| Task | What you submit | AI-use requirement | | ------------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------ | --------------------------------------------------------------------------------- | | 1. Brief the workspace assistant | A short customer email and briefing note based on the supplied facts, with the recorded briefing, checking and refinement | Use the provided workspace assistant; your recorded candidate messages are graded | | 2. Check the sales summary | Findings and corrections justified against the supplied sales figures | Your own approved AI tool is permitted; the submitted work is graded | | 3. Plan the survey analysis | A data-handling plan that follows the tool policy, protects the anonymity promise, checks the themes and explains disclosure to leadership | Your own approved AI tool is permitted; the submitted work is graded |

Free access and limits

  • The course and final exam are free. A verified email address is required to start the final.
  • You may start at most three new free exam sessions per UTC day. Resuming an existing session does not count as another start and does not reset its deadline.
  • The provided workspace assistant allows 30 turns per exam session and may be temporarily unavailable if its daily capacity is reached. Retakes follow the same limits.
  • Course practice retakes are unlimited; final-exam starts follow the daily limit above.

Scoring and the pass standard

  • Each task is scored against a published-format rubric of 4–7 analytic criteria. Each criterion is scored 0–4 against behavioral level descriptions, derived from a checklist of observable features of your submission.
  • Checklist decisions cite verbatim evidence from your submission, or your recorded candidate messages for Task 1's process criteria. Assistant messages do not establish actions you took. Claims the permitted evidence doesn't support earn no credit.
  • Criterion scores combine into a weighted composite on a 0–4 scale.
  • AI tools and the award. AI tools may be used on Tasks 2 and 3. Task 1's recorded work carries the award.
  • Provisional pass standard: Task 1 at 2.5 of 4 or higher, an overall composite of at least 2.5 of 4, and no task below 2.0. This is a provisional operational standard. No job-task analysis or empirical standard-setting study is claimed. It is reviewed against real submissions after launch.
  • A score of 2.0 does not pass. A score of 2 on every criterion gives a composite of 2.0 and does not meet the 2.5 standard. The task floor and the overall pass standard serve different purposes.

How decisions are made

  • AI grading uses three separate scoring passes on a fixed, versioned grading configuration. The passes may share a model.
  • A person reviews every fail and every flagged attempt before a decision is released. Borderline results and perfect scores get a second check by a different AI Model and go to a person unless both agree. A person also reviews a random sample of clear passes. Every credential decision is logged, and your result shows whether a person confirmed it.
  • Feedback generated by AI is labeled as AI-generated.
  • Appeals: you can appeal any result. The first appeal on each attempt is free. While Certavant has one human reviewer, an independent AI Model re-grades your work blind and a person countersigns the outcome. The outcome letter says if that person also confirmed the original decision. Appeals are decided within 5 business days.

AI-use policy for the assessment

CAI-1 is open-AI-by-design: using AI well is the skill being assessed.

You may:

  • use the AI assistant provided in the assessment workspace;
  • use another approved AI tool only for the work-product tasks (Tasks 2 and 3), following their instructions. Task 1 requires the provided workspace assistant, because its recorded conversation is graded alongside the submitted work;
  • use notes, the course materials and any reference sources.

You may not:

  • have another person do any part of the work, or share task content with others;
  • paste real confidential, client or personal data into any tool. The tasks provide all the information you need;
  • include text intended to influence the grader, such as instructions to the AI grader or hidden text. This is an integrity violation. It results in a fail and can lead to a ban and the revocation of any credential;
  • use the course tutor during the assessment. The separate workspace assistant can help with the task under its instructions; Task 1 requires that recorded workspace conversation.

The credential

  • Free, verifiable and shareable (Open Badges 3.0), with an "Add to LinkedIn profile" option.
  • Evidence-embedded: the credential can link to the work you submitted, with privacy settings you control (public, employer-link-only, or hidden).
  • Displayed as CAI-1 AI Foundations.
  • Stamped with an expiry date at issue (12–18 months; the final term is to be confirmed) and renewable by assessment.

Content freshness

Durable skills (D1–D3) form most of the course and change slowly. Tool-specific material, mainly in D4, is flagged in the content and reviewed regularly. Changes are published in the course changelog.