Resumes, self-reported skills and course certificates don't show how someone works with AI. Coodesh measures it with evidence: in hiring, each candidate's judgment in real situations from the role; on your team, the level of each person and each department, to guide development and projects.
AI reached the workplace faster than the ways of measuring who can use it. Every resume lists tools, every team says it already uses them, and HR still can't tell who makes good decisions with AI and who just copies and pastes the answer.
of U.S. job seekers used AI in their job search in the past year.
Source: Greenhouse, job seeker survey, Oct/2026of white-collar workers say they feel properly trained for AI at work.
Source: BCG, AI at Work 2026, Jun/2026of the AI users Microsoft surveyed reach its most advanced tier, Frontier Professionals.
Source: Microsoft, 2026 Work Trend Index, May/2026Knowing a tool's name doesn't set anyone apart. Judgment does: when to use it, how to prompt, how to verify and what never to paste into a chat.
The criteria are the same in both cases. What changes is what you do with the result.
Measure each candidate's judgment with AI before the hiring manager interview. Results you can compare across candidates, with identity and integrity verified.
See how it works in hiringFind out the AI fluency level of each person, team and department, and use the diagnosis for development paths, project staffing and workforce planning.
See how it works for your teamCandidates solve everyday situations from the role with AI available, as they would at work. The assessment looks at the decisions they make, not the tools they can name.
When a task is worth handing to AI, when to do it alone and how to split the work into parts AI handles well.
How they describe the goal, give context, set the output format and adjust the prompt when the result is weak.
Whether they check facts, numbers and code, catch the mistake AI delivered with confidence and know when not to trust it.
How they handle customer data, personal data and confidential information, and when the rule is not to use AI.
You decide, assessment by assessment, whether AI is allowed. When it is, the score rewards using it well. When it isn't, use outside the rule shows up as suspicious activity.
Start for freeThe diagnosis shows each person's AI fluency level and each department's maturity. With it, you decide where to invest in training, who to bring into an AI project and what you need to hire externally.
Uses it little or not at all. Doesn't yet see where AI would help in their own work.
Uses it for one-off tasks, with generic prompts, and rarely checks what comes back.
Uses it often for tasks in the role, gives context, reviews the result and knows what can't be shared.
Redesigns their own workflow with AI, combines tools and knows when AI isn't the answer.
Sets up AI for the team, creates safe-use standards and helps others improve.
Level distribution for each team, department and position, with the gap against the level you expect from each one.
Avoids AI or uses it only when asked. Usually careful with data, but leaves productivity on the table.
Uses it for everything and trusts it too quickly. Gains speed but risks passing errors along.
Uses it well on familiar tasks, with good prompts and review. Hasn't changed how they work yet.
Redesigns processes, sets up AI for others and creates usage rules for the team.
An action plan per gap, with courses from your LMS and reassessment to prove progress.
Build squads and AI projects from the levels you need and see who covers each requirement.
Hiring, redeployment and development scenarios based on what the team can already do.
A verifiable certificate for people who reach the expected level, with identity confirmed.
Each employee knows they're being assessed, why and what happens with the result, and receives their own report. We recommend sharing the goal before the cycle and using the diagnosis to plan development and projects, not as a sole criterion for dismissal.
AI fluency looks different in every function. The assessment uses situations from the role, not generic questions about tools.
Using coding assistants to move faster without losing ownership: breaking down the problem, reviewing what AI generated and answering for the result.
AI suggests a fix that passes the tests but creates a race condition. Does the person notice, explain and correct it?
Generating queries, pipelines and documentation with AI, validating everything against real data.
The query AI wrote runs without errors but duplicates records in a join. Does the person check the counts before publishing the data?
Choosing the model and approach, evaluating outputs against criteria and anticipating failures, cost and security risks.
An internal assistant starts showing one customer's data to another. What controls does the person propose and how do they test them?
Using AI in discovery, analysis and specs without trading user evidence for well-written text.
AI summarizes 40 customer interviews and points to a trend. How does the person check the summary is accurate before changing the roadmap?
Automating reports and CRM hygiene with AI without breaking business rules.
AI proposes a new lead routing rule. Does the person test the effect on territories and quotas before applying it?
Preparing accounts, proposals and follow-ups with AI without bringing the customer a fact nobody checked.
The account research AI did includes a revenue figure with no source. Does the person use it in the meeting or check it first?
Personalizing outreach at scale with messages that sound human and respect privacy laws such as GDPR.
Can the lead list, with phone numbers and emails, be pasted into a public chat to generate the messages? What does the person do?
Creating ad variations and analyzing campaigns with AI without losing control of budget and brand.
AI recommends doubling a campaign's budget based on one week of data. Does the person accept, test or push back?
Building scenarios and explaining variances with AI, checking every number against the source data.
AI explains the month's variance with a plausible cause that doesn't appear in the data. Does the person catch it?
Defining where AI can step into the close and reconciliation and where human review is mandatory.
The team wants to use AI to classify accounting entries. What controls and audit trail does the person require?
Looking up and interpreting rules with AI without treating the answer as a formal opinion.
AI cites a tax rule that recently changed. Does the person check the official source before applying it?
Automating checks and collections with AI while protecting banking data.
A vendor asks to change bank accounts in a well-written email, and AI sees no problem. Does the person?
Writing job posts, screening and preparing interviews with AI without handing it decisions about people.
AI suggests rejecting candidates based on a criterion that could discriminate. Does the person notice and correct it?
Analyzing engagement and performance with AI, taking extra care with employee data.
A manager asks to summarize the team's performance reviews in a chat tool the company hasn't approved. What does the person say?
Building learning paths and content with AI and measuring whether training changed practice.
AI generates a full course in minutes. How does the person validate the content and measure how it's applied at work?
Running people analyses with AI, with statistical rigor and privacy.
AI points to a link between remote work and attrition. Does the person treat it as the cause or investigate?
Preparing business reviews, reading risk signals and writing account plans with AI.
AI marks a customer as healthy based on product usage, but the sponsor's latest email mentions canceling. Does the person cross-check the signals?
Designing AI service flows and knowing when to hand a case to a person.
The bot gets most cases right but fails with customers in vulnerable situations. What escalation rule does the person propose?
Responding faster with AI without giving the customer wrong information.
The reply AI suggested mentions a feature the product doesn't have. Does the person send, edit or escalate?
Don't see your role? The assessment is built from the situations in your day-to-day work.
Book a demoCombine formats in a single assessment. People do what they'd do at work, and you see how they think.
Candidates code with an AI assistant, the same one for everyone, inside the assessment. The prompt history goes into the report.
See formatAI interview, by voice or chat, with follow-up questions when the answer is shallow and a score per criterion.
See formatA written response to a situation from the role, with the prompt the person would give AI and their review of what came back, graded by rubric.
See formatThe person explains and defends the decision on video, in their own words.
They draw a flow or an architecture and show where AI, human review and controls come in.
See formatScenarios with options to measure decision-making at scale, in any department.
See formatPrefer to watch live? Live Interview also assesses AI use, with criteria written by your company.
Explore Live InterviewIn an assessment about AI, integrity matters even more. You define what's allowed, and the platform records what happened.
Tab switching, pasted text, AI use outside the rule, second screen, browsing recording and webcam photos, with a severity indicator to prioritize review.
Learn moreLiveness checks and face matching before and during tests and AI interviews, to block manipulated video.
Learn moreID document and photo identification confirm the person is who they say they are.
Learn moreAn accessibility menu on every screen, with screen reader, dyslexia-friendly font and Libras (Brazilian Sign Language). The integrity system is calibrated not to penalize people who use assistive technology.
Learn moreAI scores and summarizes. The final score and the decision about each person belong to your team, with recorded criteria, in line with rules on automated decisions such as GDPR Article 22. No integrity signal fails anyone on its own, and people being assessed know from the start which features are active.
Set up the assessment in each person's language and compare teams from different countries using the same standard. If the role calls for another language, add the Language Test to the same assessment, with a level on the CEFR scale.
Hiring or team diagnosis, which roles and what level you expect from each.
Situations from the role, formats and the AI usage rule, from the Coodesh library or your own questions.
Invite by link, email or through your ATS, with proctoring and ID verification according to the risk.
ATS IntegrationA report per person and a view by department. Your team decides who moves forward and what each person will develop.
It's an assessment that measures how a person works with AI in real situations from the role: when they delegate a task to AI, how they prompt, how they check the result and how they handle sensitive data. It isn't a test about tool names. At Coodesh, it's used for hiring and for diagnosing your team's level.
AI literacy usually means understanding what AI is, how it works and what the risks are. AI fluency is putting that to work: making good decisions with AI in the day-to-day of the job. The Coodesh assessment measures the second, with practical situations from the role.
The candidate receives the invite, solves situations from the role with AI available, when the rule allows it, and can answer by text, video, code, whiteboard or in an AI Interview. The recruiter gets the score per criterion, the answers and a summary of strengths and areas of concern, along with the integrity signals.
No. AI scores and organizes the evidence using the criteria your company wrote. The final score and the decision about each person belong to your team, with the criteria recorded, in line with rules on automated decisions such as GDPR Article 22.
For development and planning: setting paths by gap, choosing who joins AI projects and planning hiring, redeployment and training. We recommend telling the team before the cycle, explaining the goal and giving each person their own report. The diagnosis shouldn't be used as a sole criterion for dismissal.
You decide, for each assessment, whether AI is allowed. When it is, AI is part of what's being measured. When it isn't, Anti-fraud and Proctoring records tab switching, pasted text, AI use outside the rule and, if enabled, the webcam. Liveness checks and face matching confirm the person answering is the right person.
For technology and business functions: product, engineering, data, sales, marketing, finance, HR and operations. The assessment uses situations from the role, and you can combine questions from the Coodesh library with your own.
English, Spanish and Portuguese. If the role requires another language, you can add the Language Test to the same assessment, with a level on the CEFR scale.
Yes. Candidate results flow back to your ATS through an API integration, and the employee diagnosis connects to your LMS to recommend courses by gap and reassess later.
Bring one role and see what the assessment would look like at your company.
AI recruiting platform to assess, interview and hire based on proven skills.
Create your account and send your first assessment today.
