Both say they’re great with AI.
Does it show in real work?
AI Literacy Assessment
Everyone’s “good with AI.”
See what that means at work.
Watch people delegate to AI, check its work, and finish the job. Get each person’s strengths and gaps, plus AI skill levels by team and role.
Candidate A
I’m good with AI.
Vague ask → No review
Overall AI levelLv1.5
Candidate B
I’m good with AI.
Plan → Ask → Review
Overall AI levelLv3.8
Fictional results for two candidates on the same task. Levels match the examples below.
Sample run · 6 scenes
Same task, two candidates.
See how they differ.
Fictional example
- Both say they’re great with AI. Does it show in real work?
- Same 25-minute task for both. Using the built-in AI, sort 300 customer inquiries and pick the top issues for a weekly meeting.
- The AI adds a cause that isn’t in the data. A uses it as is. B flags it and asks for a fix.
- Their review shows in the result. A’s draft keeps the unsupported cause and customer data. B separates facts from open questions and removes the data.
- The chat log shows how they worked. Telta AI scores the chat and final answers against set criteria.
- Levels, with the reasons. Here, A scores Lv1.5 and B Lv3.8. The report lists strengths and gaps to probe in the interview.
Contents
Reports
Every assessment comes with
individual and org reports.
Individual report · Everyone
Each person’s level, and why
Levels with evidence, peer standing, how they used AI, and strengths, gaps, and next steps. Concise for hiring; detailed, with feedback, for development.
Org report · For comparisons
AI skills by team and role
Compare level spread (Full adds AI understanding grades), team and role gaps, and AI usage. Reassess with the same version to track change.
Sample report data is fictional.
Why it matters
Don’t just track AI usage.
See how the work gets done.
What you track today
- Tools rolled out
- Accounts issued
- Usage frequency
What an assessment adds
- How they delegate to AI
- How they handle sensitive data
- Whether they check AI output
Hiring
Test the AI skills on a résumé with a real task.
Employee assessment
Compare AI levels, strengths, and gaps across teams and roles.
Safe AI use
See how people handle customer and internal data.
How we measure
What people know vs.
what they actually do.
Knowing the answer isn’t the same as doing the work. A real task shows how people check and fix AI output.
| What you need | Quiz | Self-rating | Real task |
|---|---|---|---|
| How people use AI at work | |||
| Handling sensitive data | |||
| Free of self-report bias | |||
| AI knowledge | What they know | What they think they know | What they apply |
| Delegating to AI | Limited | Self-reported only | Shown in their prompts |
| Checking AI answers | Limited | Self-reported only | Shown in their edits |
| Evidence | Correct answers | Self-scores | Chat log + final answers |
YesLimited
Quizzes test AI knowledge; self-ratings show self-perception. Telta’s tasks capture how people prompt, review, and revise, plus what they deliver.
How it works
Delegate. Review. Deliver.
We assess every step.
Every prompt counts, not just the final answer. We analyze the AI chat and the deliverable together.
Task 1
Plan
Decide what AI handles and what you check.
Task 2
Prompt & revise
Set sources and constraints, then review and refine.
Task 3
Final review
Finalize answers and pre-share checks.
See it in action
Demo · Click a step to jump to it.Tap a step to jump to it.
Fictional task, chat, and scores, for illustration only.
Click a step to jump to it.Sample results
Same task,
very different approaches.
Candidate A · “Great with AI”
Lv1.5
PlanningLv1.5
Execution & InteractionLv1.5
Review & DecisionsLv1.5
Pasted raw customer data and used the AI’s cause unchecked.
Candidate B · “Great with AI”
Lv3.8
PlanningLv4.0
Execution & InteractionLv4.0
Review & DecisionsLv3.5
Removed customer data and flagged the unsupported cause.
Fictional example · The two candidates from the video above.
Same task, same 25 minutes
Turn 300 customer inquiries into a weekly meeting brief.
The difference · Checking sources
A kept the unsupported cause. B caught it and had it marked [Needs verification].
People make the final call
Results inform. Hiring and HR managers decide.
Evidence
Scores you can trace
to the candidate’s own words.
Candidate B · Request 2
“‘Warehouse relocation’ isn’t in the materials. Mark the cause [Needs verification] and quote the messages verbatim.”
Competency
Verifying AI output
Caught a cause not in the materials and asked for a fix.
Level
Lv4
Candidate B · Task 1 answer
“Strip customer names and phone numbers before sharing messages with the AI”
Competency
Splitting human and AI work
Decided up front what the AI gets to see.
Formal name · Setting human–AI task boundaries
Level
Lv5
Fictional example. Levels come from all answers and the full chat, scored against each competency’s criteria; only key evidence is shown. See all criteria
6
Scored on six competencies
Telta AI scores the chat and final answers against each one.
Individual report
A report for every participant
Levels, evidence, strengths, gaps, and next steps.
Using reports
Know what to dig into
HR and leaders use the evidence to plan interview follow-ups and spot skills to build.
What we assess
Full framework · 6 areas · 14 competencies
AI Understanding
AI understanding and risk awareness
Do they understand AI?Full
Checks whether people understand how AI works, where it falls short, and what risks to watch for. It also covers risks such as security, copyright, and personal data, and choosing tools that fit the company’s AI policy.
Technical understandingTechnical
Explaining how AI worksUnderstanding prompt–output variabilityWhy AI answers change with the requestRecognizing AI’s limits
Risk awarenessRisk
Judging high-risk AI useSpotting situations that call for caution with AIDetecting bias in AI outputsIdentifying sensitive data
Tool understanding and selectionTool
Knowing AI tool policiesChoosing AI tools
Assessed with Full’s 16 multiple-choice questions, with a grade for each competency area.
AI Application
AI application skills
Can they get work done with AI?FullLight
Checks whether people hand work to AI with a clear purpose, review its answers, and complete the deliverable.
PlanningPlan
1Setting human–AI task boundariesDeciding what AI does and what people do2Breaking work into AI-ready tasksSplitting work into steps AI can handle
Execution and interactionExecute
3Structured promptingAsking with the goal, materials, and conditions4Iterative output refinementReviewing and improving AI answers
Review and decision-makingReview
5Verifying AI outputs6Human-accountable decision-makingPeople decide on the final result and how it’s used
Competency, area, and overall levels (Lv1–Lv5) based on the AI chat history and final answers.
How we built it
Built on AI competency research
and real job data.
What it’s built on
10 organizations
Public AI skills research from OpenAI, Anthropic, Stanford, the AI Alliance, and others
30 companies
Job data from 30 AI transformation leaders recognized by Fortune
14 competencies
8 for AI understanding and risk, 6 for AI application
Reviewed by HR Ph.D.s
HR experts reviewed the resulting framework.
One standard across roles
Sales, marketing, HR, and more, measured on one scale.
Technical track record
Telta placed first overall in TalentCLEF 2026 (HCM AI, Tasks A and B).
~25 min
Light · AI tasks
~60 min
Full · Quiz + AI tasks
Lv1~Lv5
Levels for each competency
10,000+
Teams of 10 to entire companies
Versions
Two versions,
for different goals.
Light focuses on AI use. Full adds AI knowledge. Compare below.
Best for hiring
Light
~25 minOne work scenario in three steps, covering 6 AI application competencies.
Benefits
- Quick to takeAbout 25 minutes, easy to schedule.
- Performance, not claimsSee claimed AI skills tested on a real task.
- Same task, same criteriaEveryone gets the same task, so comparisons are fair.
- Know what to ask nextA concise report flags what to ask in interviews or 1:1s.
Best for your team
Full
~60 min16 quiz questions plus 3 AI tasks, covering all 14 competencies.
Benefits
- Knowledge and practiceMeasures AI understanding, risk awareness, and hands-on skill together.
- Feedback for growthEveryone gets detailed feedback on all 14 competencies.
- Pinpoint training needsCompare teams and roles to see where training is needed.
- Track progressReassess later to see how each competency changes.
Light vs. Full
Side by side
| Light~25 min | Full~60 min | |
|---|---|---|
| Scope | AI application skills 3 areas · 6 competencies | AI understanding and risk + AI application 6 areas · 14 competencies |
| Format | One work scenario in three steps Plan → Chat with AI → Deliver and check | Part 1: 16 quiz questions Part 2: 3 AI tasks |
| Timing | ~25 min · No pausing · Self-paced across steps | ~60 min · No pausing |
| AI use | Built-in AI chat throughout | Built-in AI chat in Part 2 |
| Scoring | AI tasks: scored from the chat log and final answers. Full’s AI understanding: scored from the quiz. | |
| Results | Levels by competency, area, and overall (Lv1–Lv5) | AI understanding: grade by area AI application: Lv1–Lv5 levels |
Tasks and criteria differ by version, so compare results within one version.
Use cases
Hiring or developing your team,
get evidence on AI skills.
Hiring
Find what to probe in the interview
Light · ~25 min
Review how they worked and plan interview follow-ups.
Development
Give feedback, track growth
Full · ~60 min
Share detailed feedback, then reassess to measure change.
Organization
Spot strengths and gaps by team
Org report
Compare levels and competencies to see where to focus.
From teams of 10 to companies of 10,000+
Every participant gets a report. We can also build custom versions for your roles or industry.
Taking it
One link to start.
AI built in.
How to take it
Open the link and start. Light takes about 25 minutes, Full about 60. In Light, participants pace themselves across steps. Neither can be paused once started.
The screen
Task brief on the left, AI chat on the right. Participants choose their AI model.
Support
We plan goals, participants, and timing with you, for teams of 10 to companies of 10,000+. Everyone gets an individual report.
Custom versions
We co-design frameworks and tasks for your roles or industry.
