AI talent assessment that scores the work candidates actually do
Codility screens candidates for AI literacy in engineering and business roles on work simulations not resumes. You enable or disable the AI Assistant for each assessment, and every interaction a candidate has with it is captured as reviewable AI activity beside the submitted output.
Trusted by GitHub, SpaceX, Tesla, EY, LSEG, SAP, Barclays and Citi
Trusted by GitHub, SpaceX, Tesla, EY, LSEG, SAP, Barclays, Citi.
Codility is rated 4.6 of 5 on G2 and ranked #1 for Enterprise Technical Skills Screening. Rankings come from verified customer reviews, not paid placements, and are updated quarterly.
Your screening stage was built for a world without AI
There are four ways teams screen candidates today, all of which lead to poor decisions.
01
Recruiters reading every resume by hand
A resume gives recruiters limited evidence of how someone works. AI-written applications make that judgment harder. A strong candidate can be overlooked because their resume does not show their skills clearly.
02
AI resume screeners and AI-scored video interviews
These tools help teams review applications faster, but a resume or video response gives limited evidence of what a candidate can produce. If a candidate questions the result, your team still needs to explain how they were assessed.
03
Multiple-choice skills tests
A knowledge score tells you who studied, or checked online. Fabric found cheating behavior in 38.5% of candidates across more than 19,000 technical interviews, and 61% of the candidates it identified scored above the passing threshold, so they would have advanced undetected.
04
Ad hoc interviews
When hiring managers use different questions and evaluation criteria, candidate results are hard to compare. Without a shared record, decisions can depend too heavily on what each interviewer remembers.
Karat’s 2025-2026 AI Workforce Transformation Report found that confidence among engineering leaders that the right candidates get offers dropped from 68% to 47% in a single year, and that 71% of US engineering leaders now find AI-era skills hard to assess.
Give candidates real work, let them use AI, and score what they produce
AI talent assessment means evaluating a candidate on work they produce during the test, with AI available in the assessment the way it is available in the job. AI candidate screening covers two different products: AI resume screening uses a model to rank applications, and is not defensible. AI talent assessment scores the work a candidate produces, allowing you to correctly judge capability.
Codility provides assessments that score candidates real abilities
- Candidates work through a task simulation drawn from a library of more than 1,300 validated tasks, plus more than 50 validated business tasks for roles outside engineering.
- The AI Assistant is enabled or disabled per assessment, and every interaction a candidate has with it is captured as reviewable AI activity.
- The score is based on the work submitted.
- Codility scores the work and shows what produced the score.
- Codility doesn’t reject candidates, build your shortlist, or decide who gets hired.
- That decision stays with you, and the record of how you made it stays with the candidate report.
Screen every applicant on a realistic work simulation
- Codility Screen sends one task-based assessment to everyone who applies, for engineering roles and for business roles.
- Codility scores each submission against the task’s test cases.
- Every task is designed by assessment engineers, reviewed by occupational psychologists, and tested for AI solvability before it reaches a candidate.
- Codility Screen connects with 20+ platforms, including Greenhouse, Lever, Ashby, Workday, SAP and SmartRecruiters, so the assessment sits inside the pipeline you already run.
- Across more than 1,700 engineering evaluations in Codility’s scoring validation study, Codility’s scores lined up with how managers rated code quality more than 90% of the time.
Your recruiters decide who moves forward based on the work each candidate submits, with the score, the test cases and the AI activity clearly visible.
Let candidates use AI and inspect how they used it
- Companies hiring engineers today have had two bad options: ban AI in the assessment and test for a job nobody does any more, or try to detect AI afterward.
- Cuellar Argotty and Manrique tested seven AI-code detection tools across 1,644 code samples and concluded that reliable AI-code detection does not currently exist.
- Codility takes the other approach. The AI Assistant is enabled or disabled per assessment, and every candidate interaction with it is captured as reviewable AI activity.
- Integrity signals are provided alongside the AI activity: behavioral signals such as paste events, focus loss and timing anomalies; cross-candidate similarity checks; identity and network checks
- The signals roll up into a single Integrity Risk level on every candidate report, with the contributing signals open for inspection.
Reviewers see how someone worked with AI, which is what the job requires. You are not wondering afterward whether the candidate used it.
Every technical interview has a record you can defend
- In Codility Interview, candidates work in a shared VS Code workspace with a terminal and autocomplete. Interviewers can also cover system design and use a whiteboard in the same session.
- Interviewers can bring their own tasks and score against a shared rubric.
- Every session produces rubric scores, a transcript and session playback.
- Underneath both stages, screening and interview, the scoring is deterministic.
- The same response always earns the same score, and neither AI nor machine learning makes an automated employment decision.
- Codility runs EEOC-aligned adverse impact analyses, including four-fifths-rule testing, and the platform is designed to support NYC Local Law 144.
- SOC 2 Type II audited, ISO 27001 certified, GDPR and CCPA compliant, WCAG 2.2 AA accessible, with EU or US data hosting chosen at contract time.
- Candidate and employee data is never used to train AI models.
When someone asks how the decision was made, you open the session record. The candidate, the auditor and your legal team all get the same answer.
Deterministic and realistic
The deterministic score Codility provides comes from the work produced by the candidate
By contrast, tools that use an LLM to grade a the submission can return different scores for the same piece of outputwork. Some tools let a recruiter edit the score afterward, but a score you can quietly override is a score you can’t defend six months later.
Candidates are given access to AI for realistic testing, and the use is recorded
The AI Assistant is switched on or off per assessment, and every interaction is captured as reviewable AI activity. Reviewers can see exactly how candidates use AI, with strong integrity signals to ensure they know the work is the candidates.
Codility publishes the full candidate dataset.
“
“With Codility, our teams ran 750 candidate tests over 90 days, saving 2,200 hours of interview time. That kind of productivity is like gaining time to launch an entirely new product or enter a new vertical.”
“
“Codility helps us remove bias by ensuring a consistent, equal, and fair process for every candidate at LiveRamp.”
Codility publishes the full candidate dataset, including the numbers that do not flatter.
- 91% say the assessment content is fair.
- 87% rate the experience Good or Excellent.
- Fairness sits above 83% across every demographic group measured, and women rate content fairness higher than men do.
- Even candidates who score in the bottom quartile rate the experience Good or Excellent 60% of the time.
Source: Codility post-assessment candidate feedback survey, 48,000+ responses, Oct 2025 to Apr 2026.
The same screening standard for roles outside engineering
Most teams start with engineering, because that’s where the volume is. The same platform provides screening for marketing, sales, finance and operations candidates on work simulations built for those roles, with the same scoring and the same reporting.
You can also assess how your current employees work with AI.
Frequently asked questions
What is AI talent assessment?
AI talent assessment is the practice of evaluating candidates on work they produce with AI available during the assessment, and scoring the output. The candidate completes a role-relevant task simulation rather than answering questions about it, and the hiring team gets a score along with the work behind it. It answers the question “can this person do the job the way the job is now done?”
Is this AI screening the candidate, or assessing what the candidate can do with AI?
How is this different from an AI video interview?
Can candidates use AI during the assessment?
How do you keep the screening decision defensible?
Which roles can you screen?
Does it work with our ATS?
How long before we are running?
AI talent assessment for teams who have to defend the decision
Start your free trial
Send your first work-simulation assessment on one of your open roles
SOC 2 Type II · ISO 27001 · GDPR and CCPA · EU or US data hosting