Certification Course · Professional Track

Certified AI Practitioner (CAIP).

The foundation credential for professionals applying AI in their daily work. Delivered through Authorized Training Partners, assessed independently by AICA, and issued with a verifiable digital badge.

Exam Specification

The Specification, on the Record.

Get Certified Talk to Us Download the Candidate Blueprint

Certifying a team instead? Explore workforce certification →

CredentialCertified AI Practitioner (CAIP)
TrackFoundation credential (entry to the Professional Track)
Assessment formatProctored examination (100% of the assessment weighting)
Contact hours16 contact hours across two days
Certification feeUSD 1,600
Validity3 years from award date
Renewal20 CPD hours per 3-year cycle, logged with AICA
Retake policyReattempt after a 14 day waiting period, up to 3 attempts in any 12 months
DeliveryThrough Authorized Training Partners, online or center-based
VerificationCryptographic registry entry, QR-verifiable digital badge, Open Badges 3.0

Exam duration and question counts are set in the Candidate Handbook and are not published here. The full proctoring, retake, appeals, renewal and revocation terms are on the exam policies page.

Course Overview

The Baseline for the AI-Ready Workforce.

The CAIP certification course prepares professionals for the Certified AI Practitioner credential, AICA's foundation certification for people applying AI in their daily work. It validates core AI literacy and the practical, responsible use of AI tools in real business workflows: the baseline competency for the AI-ready workforce.

AI tools have arrived in every function, but competence has not arrived with them. Organizations need a way to distinguish staff who use AI well, with judgment about accuracy, confidentiality and fit for the task, from staff who simply use it. The CAIP provides that benchmark: an independently assessed standard for what every professional should know and be able to do with AI, whatever their role.

The CAIP is the entry point to the Professional Track in the AICA certification portfolio, and the natural first step toward specialized credentials such as the CAAP and CAIGP. Like every AICA credential, it follows a governed process in which standards, training and assessment are deliberately separated. The full model is set out on the How It Works page.

Who It Is For

  • Professionals in any function using AI tools in daily work
  • Managers and team leads setting the standard for AI use in their teams
  • Analysts, marketers, operations and HR staff building applied AI skills
  • Career changers establishing a verified foundation in AI

The Mandate It Validates

Everyday professional competence with AI: understanding how the tools work, applying them productively in real workflows, and using them responsibly and securely.

Competency Domains

Six Domains. One Standard of Competence.

The CAIP competency framework is built on six domains. Certification confirms demonstrated capability in each, assessed against predefined benchmarks rather than attendance.

01

AI & Machine Learning Fundamentals

Certified practitioners can explain, in plain language, what AI and machine learning systems do, where they are strong and where they fail. They can set realistic expectations for what a tool will and will not deliver on a given task.

02

Prompt & Context Engineering

Holders can write prompts that reliably produce useful output: providing context, constraints and examples, and iterating deliberately when results miss. They can adapt their approach across different tools and models.

03

Applied AI in Business Workflows

Holders can identify where AI genuinely improves a workflow, then build it into their routine: drafting, analysis, summarization, research and review. They can measure whether the change actually saves time or lifts quality.

04

Data Fundamentals & Quality

Holders can judge whether the data behind a task is fit for purpose: complete, current and representative. They can spot quality problems that will mislead an AI tool, and know when a human check is non-negotiable.

05

Responsible & Secure AI Use

Holders can use AI within the rules: protecting confidential and personal data, respecting organizational policy, verifying outputs before relying on them, and disclosing AI assistance where it matters.

06

AI Tool Evaluation & Selection

Holders can compare AI tools against a real requirement rather than a trend: capability, cost, data handling and vendor credibility. They can recommend a tool with reasons, and recognize when no tool is the right answer.

Course Curriculum

The Full CAIP Curriculum, Published in the Open.

The complete curriculum is published here so that candidates, employers and Authorized Training Partners can see exactly what the credential covers. Training partners deliver to this curriculum; assessment and the certification decision remain with AICA.

Duration
16 contact hours
Delivered across two consecutive or split days.
Format
Two-day workshop
Instructor-led sessions with guided practice on real workplace tasks.
Assessment
Proctored examination
Carries 100% of the assessment weighting. Administered independently of training.
Entry Requirements
None
No prior technical background is required, and the course assumes none. Every unit starts from the work people already do.
Designed For
Working professionals
Any professional applying AI in daily work, in any function.
Course-Level Learning Objectives

What Completion Certifies.

On completion of the CAIP certification course, candidates are able to:

  1. Explain, in plain language, what AI, machine learning, generative AI and agentic AI systems do, where each is strong and where each fails, and set realistic expectations for a given task.Traces to LU1
  2. Construct prompts that supply the context, constraints and examples a tool needs, and iterate deliberately when a first result misses the mark.Traces to LU2
  3. Apply AI tools to everyday workflows such as drafting, analysis, summarization, research and review, with human checks at the points where an error would carry cost.Traces to LU3
  4. Evaluate whether the data behind a task is fit for purpose, and identify quality problems that would mislead an AI tool or the person relying on it.Traces to LU4
  5. Protect confidential, personal and commercially sensitive information when working with AI, including recognizing what must never be entered into a public tool.Traces to LU5
  6. Verify AI output against reliable sources before acting on it, and disclose AI assistance where colleagues, clients or organizational policy would reasonably expect to know.Traces to LU5
  7. Evaluate an AI tool against a genuine requirement, weighing capability, cost, data handling and vendor credibility, and recommend adoption, rejection or deferral with reasons.Traces to LU6
Learning Units

Six Units. Sixteen Contact Hours.

The six learning units follow the six CAIP competency domains in order. Each unit states its numbered learning outcomes with the assessment criteria that evidence competent performance, its delivery method, then the attitudes, skills and knowledge a competent practitioner demonstrates, following the Attitude, Skills, Knowledge (A.S.K.) convention.

LU1

AI & Machine Learning Fundamentals

2.5 contact hours · Day One

What AI systems actually do when they produce a result, and where they predictably fail. This unit replaces both hype and dismissal with a working mental model that holds up in daily use.

DeliveryInteractive briefing with worked demonstrations, followed by a guided exercise classifying the AI tools candidates already use at work.

Learning Outcomes

Traces to CLO1
  • LO 1.1Distinguish AI, machine learning, generative AI and agentic AI, and describe in plain language what each is doing when it produces a result
    Assessment criteria
    • Sorts a set of familiar workplace tools, such as a chat assistant, a transcription service and a recommendation engine, into the correct category of AI
    • Explains to a colleague, without jargon, what a generative tool is doing when it drafts a document
  • LO 1.2Explain common failure modes, including hallucination, staleness and bias, and predict where a given tool is likely to be unreliable
    Assessment criteria
    • Names the likely failure mode when shown an AI answer that is fluent but wrong, such as a fabricated citation or an outdated figure
    • Points out where a given tool is likely to be unreliable on a specific workplace task, and gives a reason grounded in how the tool works
  • LO 1.3Set realistic expectations for what an AI tool will and will not deliver on a specific task
    Assessment criteria
    • States, for a specific task, what the tool can be expected to deliver and what it cannot know unless it is told
    • Corrects an overclaimed or dismissive statement about an AI tool with a realistic account of its strengths and limits
A.S.K. DimensionStatements
Attitude
  • Curiosity about how a tool produces its output, rather than treating the output as either magic or nonsense
  • Calibrated realism: resists both overclaiming and dismissal when judging what a tool can do
Skills
  • Describe, without jargon, how a machine learning model learns from data and how a generative model produces output
  • Classify everyday tools, such as chat assistants, recommendation engines, transcription services and agent-style automations, by the kind of AI at work
  • Predict the likely failure points of a tool on a given task, and state what the tool cannot know
Knowledge
  • The distinctions between rule-based software, machine learning, generative AI and agentic AI
  • How models are trained on data, and why training data shapes behavior, including knowledge cutoffs and bias
  • Common failure modes: hallucination, confident error, staleness, bias, and sensitivity to phrasing
  • What a model does and does not have access to: no private context unless provided, no live information unless connected
LU2

Prompt & Context Engineering

3 contact hours · Day One

Getting reliably useful output from AI tools. The unit treats a prompt as a brief to a capable but uninformed colleague, and builds the habit of diagnosing weak results instead of retrying at random.

DeliveryGuided hands-on practice at the keyboard: candidates write, test and revise prompts on their own workplace tasks, with instructor feedback on each iteration.

Learning Outcomes

Traces to CLO2
  • LO 2.1Construct prompts that state the task, audience, format and constraints, and supply the context the tool needs to do the work well
    Assessment criteria
    • Writes a prompt for a real workplace task that states the role, task, audience, format and constraints, and attaches the context the tool needs
    • Produces usable first-pass output on a routine drafting task by briefing the tool as they would brief a capable colleague
  • LO 2.2Diagnose a weak result, identify what was missing from the prompt, and revise deliberately
    Assessment criteria
    • Identifies which element of the prompt caused a weak result, such as missing context or an unstated format, and revises that element rather than retrying at random
    • Improves a poor output through deliberate follow-up instructions, including asking the tool to critique its own draft
  • LO 2.3Adapt prompting technique across different tools and models
    Assessment criteria
    • Runs the same task brief on two different tools and describes where the behavior differs and what that means for the task
    • Adjusts a prompt that worked in one tool so that it performs in another, rather than abandoning the task when results change
A.S.K. DimensionStatements
Attitude
  • Precision: treats a prompt as a proper brief, not a search query
  • Persistence with method: a poor first answer triggers diagnosis, not abandonment
Skills
  • Write a prompt that specifies role, task, audience, format, constraints and success criteria
  • Provide context effectively: background material, examples of good output, definitions of terms the tool cannot infer
  • Diagnose a weak output and identify which element of the prompt to change
  • Refine iteratively, including follow-up instructions and asking the tool to critique its own draft
  • Adapt the same task brief across different tools and note where behavior differs
Knowledge
  • Why context determines output quality: what a model can and cannot infer from a short instruction
  • The core prompt elements: role, task, context, constraints, examples and output format
  • Everyday prompting techniques: step-by-step instructions, worked examples, and requesting alternatives
  • The limits of prompting: no phrasing compensates for missing information or a capability the tool does not have
LU3

Applied AI in Business Workflows

2.5 contact hours · Day One

Moving from occasional use to a working routine. Candidates map their own workflows, place AI where it genuinely helps, position human review where an error would carry cost, and measure the result.

DeliveryScenario exercise: candidates map one of their own workflows, place AI and review steps within it, and present the result for group critique.

Learning Outcomes

Traces to CLO3
  • LO 3.1Identify the tasks in a workflow where AI genuinely improves speed or quality, and the tasks where it does not
    Assessment criteria
    • Marks, on a map of their own workflow, the steps where AI assistance would save time or lift quality and the steps where it would not
    • Gives a reason for rejecting AI on a step where the risk or the effort outweighs the benefit
  • LO 3.2Build an AI step into a routine workflow, with a human review point placed where accuracy matters
    Assessment criteria
    • Builds an AI step into a routine task and places a human review point at each stage where an error would carry cost
    • States what the reviewer checks at each review point, rather than relying on a general instruction to look the work over
  • LO 3.3Measure whether the change actually saved time or lifted quality
    Assessment criteria
    • Compares time taken and revisions needed before and after introducing an AI step, using records rather than impressions
    • Decides from the comparison whether to keep, adjust or drop the AI step, and can defend that decision
A.S.K. DimensionStatements
Attitude
  • Outcome focus: adopts AI where it measurably helps, not because it is new
  • Ownership: the professional, not the tool, remains accountable for the final product
Skills
  • Map a personal workflow and mark the steps suitable for AI assistance
  • Use AI for drafting, summarization, analysis, research and review tasks with appropriate oversight
  • Place a human review step where an error would carry cost, and define what the reviewer checks
  • Compare before and after: time taken, revisions needed, quality of the result
Knowledge
  • Task types where current AI tools perform well, and task types where they perform poorly
  • The human-in-the-loop principle, and where review points belong in a workflow
  • Practical workflow patterns: first-draft generation, summarizing long material, structured extraction, research scoping
  • Simple measures of workflow improvement: time saved, error rates, rework avoided
LU4

Data Fundamentals & Quality

2 contact hours · Day Two

An AI tool amplifies the data it is given. This unit equips candidates to judge whether that data is fit for purpose before trusting anything built on it.

DeliveryInteractive briefing followed by guided hands-on practice assessing a flawed sample dataset against the quality dimensions.

Learning Outcomes

Traces to CLO4
  • LO 4.1Judge whether the data behind a task is fit for purpose, using recognized quality dimensions
    Assessment criteria
    • Assesses a dataset or source document against accuracy, completeness, currency, consistency and representativeness, and states which dimensions fall short
    • Explains how a specific quality problem in the data would show up in the AI output built on it
  • LO 4.2Identify data problems that will mislead an AI tool or its user, and decide when a human check is non-negotiable
    Assessment criteria
    • Spots duplicates, gaps, outdated records and unrepresentative samples in a working dataset before the data is used
    • Decides when the state of the data demands a human check before results are acted on, and states what that check must confirm
A.S.K. DimensionStatements
Attitude
  • Skepticism proportionate to stakes: the higher the consequence, the harder the look at the data
  • Care with provenance: asks where the data came from before asking what it says
Skills
  • Assess a dataset or source document against the quality dimensions: accuracy, completeness, currency, consistency and representativeness
  • Spot common quality problems: duplicates, gaps, outdated records, unrepresentative samples and mislabelled fields
  • Trace a wrong AI output back to its inputs and identify whether the fault lies in the data
  • Decide when data quality demands a human check before results are used
Knowledge
  • The main data quality dimensions, and what each looks like in everyday business data
  • How poor input data degrades AI output, and why the degradation is often invisible in a fluent answer
  • The difference between structured and unstructured data, and which tools handle each
  • Why representativeness matters: results reflect the data the tool saw, not the world as it is
LU5

Responsible & Secure AI Use

3.5 contact hours · Day Two

The unit with the sharpest edges: what must never be entered into a public tool, how to verify output before relying on it, and when to disclose AI assistance. This is the conduct the credential most directly vouches for.

DeliveryScenario exercise on realistic confidentiality and disclosure cases, followed by group critique of AI output containing planted errors.

Learning Outcomes

Traces to CLO5 and CLO6
  • LO 5.1Protect confidential, personal and commercially sensitive data when using AI tools, including recognizing what must never be entered into a public tool
    Assessment criteria
    • Identifies which categories of workplace information must never be entered into a public AI tool and selects a compliant alternative
    • Classifies information as public, internal, confidential or personal before entering it into any AI tool
  • LO 5.2Verify AI output before relying on it, and disclose AI assistance where colleagues, clients or organizational policy expect it
    Assessment criteria
    • Checks factual claims, figures and citations in an AI draft against reliable sources before the draft is sent or acted on
    • Recognizes the situations where colleagues, clients or policy would reasonably expect AI assistance to be disclosed, and words the disclosure appropriately
  • LO 5.3Apply organizational policy and basic data protection obligations to daily AI use
    Assessment criteria
    • Applies the organization's AI use policy to a day-to-day case, and defaults to caution where no policy exists
    • Recognizes when material about to be entered into a tool contains personal data protected by law, and withholds it
A.S.K. DimensionStatements
Attitude
  • Verify-before-trust habit: no AI output is repeated, sent or acted on unchecked when the outcome matters
  • Data protection mindfulness: pauses before pasting, and treats every prompt field as a potential disclosure
  • Openness: discloses AI assistance rather than presenting AI output as unaided work
Skills
  • Classify information before entering it into a tool: public, internal, confidential or personal
  • Recognize what must never be pasted into a public AI tool: personal data, credentials, client-confidential material, unreleased financials and proprietary information
  • Verify factual claims, figures and citations in AI output against reliable sources
  • Apply the organization's AI use policy, and default to caution where no policy exists
  • Disclose AI assistance appropriately in documents and communications
Knowledge
  • How public AI tools and enterprise deployments differ in the way data is handled and retained
  • The categories of data that carry legal or contractual protection, including personal data under data protection law
  • The failure modes that make verification necessary: fabricated citations and plausible but wrong figures
  • When and how to disclose AI assistance, and why undisclosed use damages trust
LU6

AI Tool Evaluation & Selection

2.5 contact hours · Day Two

Choosing tools against a real requirement rather than a trend. Candidates finish able to run a structured trial, read a vendor's data handling terms, and put a recommendation in writing.

DeliveryGuided hands-on evaluation: candidates trial a tool against a stated requirement on safe work samples, then present a written recommendation for group critique.

Learning Outcomes

Traces to CLO7
  • LO 6.1Evaluate an AI tool against a real requirement using a structured set of criteria
    Assessment criteria
    • Defines the requirement a tool must meet before looking at candidates, and compares tools on capability, cost, data handling and vendor credibility
    • Reads a vendor's data handling terms and identifies deal-breakers, such as training on customer data or indefinite retention
  • LO 6.2Recommend adoption, rejection or deferral with reasons, and recognize when no tool is the right answer
    Assessment criteria
    • Writes a short recommendation to adopt, reject or defer a tool, with reasons an informed colleague could follow
    • Recognizes the cases where no tool is the right answer and says so, rather than recommending the least bad option
A.S.K. DimensionStatements
Attitude
  • Requirement first: starts from the problem to be solved, not from the tool on offer
  • Skepticism toward vendor claims: expects evidence, runs a trial, and reads what happens to the data
Skills
  • Define the requirement a tool must meet before looking at candidates
  • Compare tools on capability, cost, data handling, security posture and vendor credibility
  • Run a structured trial on realistic but safe work samples, and record the results
  • Read a vendor's data handling terms well enough to spot deal-breakers, such as training on customer data or indefinite retention
  • Write a short recommendation: adopt, reject or defer, with reasons
Knowledge
  • The evaluation criteria that matter: capability fit, total cost, data handling, support and fit with existing workflows
  • The questions to ask about data handling: where data is stored, whether it is used for training, and how long it is retained
  • Warning signs in vendor claims: no trial offered, benchmark claims without method, and guarantees of outcomes
  • Why no tool is sometimes the right recommendation, and what adopting the wrong one costs
Assessment Blueprint

How the Examination Maps to the Units.

The proctored examination is the sole assessment instrument and every learning unit is examinable. Questions are scenario based, testing applied judgment rather than recall. Emphasis is indicative and describes the relative weight given to each unit in the examination as a whole.

Learning UnitCompetency AssessedCourse-Level ObjectiveAssessment MethodIndicative Emphasis
LU1 · AI & Machine Learning FundamentalsExplaining what AI systems do and where they fail; setting realistic expectations for a given taskCLO1Proctored examinationCore emphasis
LU2 · Prompt & Context EngineeringConstructing and refining prompts; diagnosing weak output and revising with methodCLO2Proctored examinationCore emphasis
LU3 · Applied AI in Business WorkflowsPlacing AI and human review points in a workflow; judging where AI genuinely helpsCLO3Proctored examinationCore emphasis
LU4 · Data Fundamentals & QualityJudging whether data is fit for purpose; spotting quality problems that mislead a tool or its userCLO4Proctored examinationSupporting emphasis
LU5 · Responsible & Secure AI UseProtecting confidential and personal data; verifying output and disclosing AI assistanceCLO5, CLO6Proctored examinationCore emphasis
LU6 · AI Tool Evaluation & SelectionEvaluating a tool against a real requirement; recommending adoption, rejection or deferral with reasonsCLO7Proctored examinationSupporting emphasis

Emphasis descriptors are qualitative. AICA does not publish fixed question distributions; each examination is assembled against the competency benchmarks for all six domains, and certification decisions are made independently by the AICA Certification and Standards Authority.

Credential Terms

Validity, Renewal and Conduct.

The terms below govern the CAIP credential once awarded. They apply to every holder, in every market, under the same governed process.

TermProvision
Credential validityThree years from the date of award
RenewalThrough Continuing Professional Development: 20 CPD hours per three-year cycle, logged with AICA
RetakeUnsuccessful candidates may reattempt after a 14 day waiting period, up to a maximum of three attempts in any 12 months
AppealsCertification decisions may be appealed to AICA's Certification and Standards Authority
ProctoringThe examination is proctored, online or center-based, through Authorized Training Partners
ConductCertification requires agreement to the AICA Code of Professional Conduct

Curriculum Standard v1.0. Published 10 July 2026. Reviewed annually by the AICA Certification and Standards Authority.

Assessment & Credential

Independently Assessed. Verifiably Credentialed.

01

Assessment Format

The CAIP is assessed through a proctored examination, designed to test practical understanding and applied judgment rather than memorized definitions.

02

Delivery Through Authorized Training Partners

Preparation is delivered worldwide by AICA Authorized Training Partners: approved organizations that teach to the AICA competency framework under consistent quality requirements.

03

Independent Certification Decision

Certification decisions are made by the AICA Certification and Standards Authority, separate from training delivery. The governed, seven-stage process is set out on the How It Works page.

04

Digital Badge & Registry

Successful candidates receive the official CAIP digital badge with a unique credential identifier, recorded in the AICA verification registry. Any employer can verify the credential against its live registry record.

Frequently Asked Questions

CAIP Course FAQs.

What is the CAIP certification course?
The CAIP certification course prepares professionals for the Certified AI Practitioner credential, awarded by the Artificial Intelligence Certification Authority (AICA). It covers AI and machine learning fundamentals, prompt and context engineering, applied AI in business workflows, data fundamentals, responsible and secure AI use, and AI tool evaluation.
Who should pursue the CAIP?
The CAIP is designed for professionals in any function who apply AI in their daily work, including analysts, marketers, operations and HR staff, managers and team leads. No technical background is required.
How is the CAIP assessed?
The CAIP is assessed through a proctored examination. All certification decisions are made independently by the AICA Certification and Standards Authority, separate from training delivery.
Is the CAIP a stepping stone to other AICA certifications?
Yes. The CAIP establishes the foundation of AI literacy on which the Professional Track credentials build. Practitioners who want to specialize can progress to the Certified Agentic AI Professional (CAAP) or the Certified AI Governance Professional (CAIGP).
Related Certifications

Where This Credential Sits.

Professional Track

Certified AI Governance Professional

CAIGP

For practitioners who implement AI governance day to day.

View Certification
Professional Track

Certified Agentic AI Professional

CAAP

For practitioners who design, build and operate agentic AI systems.

View Certification

The CAIP is the foundation of the ladder and the entry point to the Professional and Executive tracks.

Ready to Earn the CAIP?

The Certified AI Practitioner program is delivered worldwide through AICA Authorized Training Partners.

Get Certified
Talk to Us