AI Skills Matrix Template 2026: 5 Levels and 6 Skill Domains for HR

An AI skills matrix template gives HR and managers a shared view of who can do what with AI — and at what level. The tool-agnostic framework works with ChatGPT, Copilot or embedded AI features in Office and HR systems. It makes training more targeted, promotions fairer, and turns the AI literacy obligations under Article 4 of the EU AI Act (effective February 2, 2025) into something you can systematically plan and document.

Skill domain	Aware	Beginner	Practitioner	Power User	Champion
AI fundamentals & concepts	Recognises terms like ChatGPT, generative AI and machine learning. Can explain in simple words why colleagues use AI for drafts, research or translations.	Uses simple prompts to answer questions or summarise texts. Follows team rules when reminded and asks before using AI with internal data.	Explains strengths and limits of AI to others (hallucinations, bias, training data). Uses AI weekly for concrete tasks and checks output critically.	Identifies new AI use cases in their area, compares AI vs. non-AI options and documents impact such as time saved or error rates reduced.	Sets AI vision and priorities for their function, connects them to strategy and KPIs and communicates a realistic view of benefits and limits.
Prompting & workflow design	Understands that clear instructions change AI output quality. Uses predefined example prompts from colleagues without major adaptation.	Writes short prompts with basic context (role, task, language). Iterates once or twice to improve results but rarely documents what worked.	Designs multi-step prompt workflows (e.g. "draft → critique → improve") for recurring tasks. Saves and reuses prompts and shares them in a team space.	Builds end-to-end AI-assisted processes such as candidate communication flows or monthly report generation. Tests variations and standardises best prompts.	Defines organisation-wide prompt patterns, templates and naming standards. Coaches teams on designing robust workflows and evaluates new AI features.
Data literacy & privacy (incl. GDPR)	Knows that AI relies on data and that GDPR and company policies apply. Avoids sharing obviously sensitive information in public tools when reminded.	Consistently avoids personal data in public AI tools. Uses simple checks (spot-checks, second pair of eyes) to see if AI output looks plausible.	Applies concepts like data minimisation and anonymisation. Prepares clean, anonymised inputs, documents sources and flags suspicious outputs.	Designs data flows and documentation for AI use cases in their area. Works with the data protection officer and IT to define retention and access rights.	Leads AI data governance across units. Reviews high-risk use cases, ensures audits and DPIAs are in place and proactively involves the Datenschutzbeauftragte.
Tool proficiency (ChatGPT, Copilot, Office, HR tools)	Is aware that tools like ChatGPT, Copilot or AI features in HRIS exist. Has watched a demo but uses them only when someone sets everything up.	Uses AI features in daily tools with guidance: drafts emails in Outlook, uses Excel AI formulas. Needs support to troubleshoot errors or logins.	Works independently with approved AI tools. Builds small automations (e.g. standard response templates, 1:1 note summaries) and helps colleagues with simple issues.	Customises tools for team needs (prompt libraries, HR templates, Copilot views). Creates short how-to guides and runs internal micro-trainings.	Evaluates new AI tools with IT and HR. Decides which tools to pilot, coordinates rollouts and ensures documentation, training and support exist.
Collaboration, communication & change	Is open-minded about AI but unsure how it affects their job. Listens to others' experiences and joins AI demos without leading them.	Shares simple AI tips with colleagues. Asks for feedback on AI-generated drafts and adapts when others raise concerns.	Leads small peer sessions to show concrete AI workflows for HR, leadership or IC tasks. Addresses fears and helps create psychological safety for experiments.	Drives cross-team AI initiatives such as an enablement community or prompt library. Removes blockers like unclear rules or missing tool access.	Integrates AI topics into company rituals (offsites, leadership meetings, works council dialogues). Connects AI use with culture, wellbeing and job design.
Governance, risk & ethics	Understands that AI can create legal, ethical and reputational risks. Knows the company has AI rules and where to find them, but still needs reminders.	Follows do's and don'ts (no confidential data in public tools, no automated hiring decisions). Escalates unclear cases instead of guessing.	Identifies potential risks in AI outputs (e.g. biased wording in job ads) and suggests mitigations such as human review steps or alternative prompts.	Writes or updates team-level AI guidelines with HR, Legal and IT. Trains others on safe usage and keeps logs of important AI-supported decisions.	Owns the AI governance framework for the organisation. Aligns with emerging regulation (e.g. EU AI Act), coordinates Betriebsrat consultations and monitors risks.

What this AI skills matrix template is — and what it does

This AI skills matrix template is a behaviour-based framework that describes AI-related competencies from basic awareness to strategic leadership. It is deliberately tool-agnostic: whether an organisation uses ChatGPT, Microsoft Copilot, or embedded AI features in HRIS, recruiting or collaboration tools makes no difference. What matters are observable behaviours.

HR, managers and employees use the matrix to align career paths, promotion criteria, performance conversations, development plans and peer feedback around a shared language. Translating role requirements into concrete, verifiable behavioural anchors makes decisions fairer — and easier to document and justify.

The 5 skill levels: Aware to Champion

Each level describes how independently someone works with AI, how complex their use cases are and how strong their multiplier effect is. The first two levels involve executing defined AI tasks with guidance. From Practitioner onward, someone takes independent ownership of workflows. Power Users and Champions design, enable and steer.

Level	Short definition	Typical scope
Aware	Understands basic concepts, observes demos, uses AI only with close guidance and predefined prompts.	Own tasks, closely guided
Beginner	Executes simple AI tasks with support, follows rules, starts forming own prompts for daily work.	Recurring routine tasks
Practitioner	Uses AI independently for core tasks, builds stable workflows and improves them through feedback and metrics.	Own projects and processes
Power User	Builds robust AI processes, enables others, measures business impact and influences tool and process choices.	Team and cross-function use cases
Champion	Shapes vision, governance and culture, connects AI use to strategy, compliance and organisation-wide change.	Multiple functions, company level

A practical example: in an HR team, a Beginner uses ChatGPT to draft job ads from a template. A Practitioner colleague has built a complete, GDPR-safe sourcing workflow with anonymised profiles. The Champion role negotiates binding AI guidelines for recruiting with the works council and Legal, and ensures they are applied across all business units.

The 6 skill domains in detail

The six domains cover what knowledge workers and managers need to use AI safely and productively. They range from basic understanding of AI concepts to governance, risk and change leadership. When defining these domains, involve IT, Legal or the data protection officer, the works council and business leaders. This prevents building a pure "tech toy" matrix with no connection to real role profiles and performance processes.

AI fundamentals & concepts: from "can explain ChatGPT in simple words" to "sets priorities and vision for AI use". Baseline knowledge for everyone.
Prompting & workflow design: from one-off prompts to documented, multi-step workflows that make core processes stable and reproducible.
Data literacy & privacy: from "knows GDPR exists" to "owns AI data governance with audits and DPIAs". Particularly critical in DACH.
Tool proficiency: from watching demos to piloting, configuring and rolling out tools company-wide — together with IT and HR.
Collaboration & change: from sharing tips to designing AI enablement and psychological safety organisation-wide.
Governance, risk & ethics: from following basic rules to defining the company-wide governance framework and aligning it with regulatory requirements.

Linking the skill domains to existing skill management processes prevents the matrix from running as a parallel silo. The goal is integration into performance cycles and internal mobility processes.

Rating scale and evidence

A robust rating rests on observable behaviour backed by concrete evidence — not impressions. Two recruiters both use AI for job ads. One copies a prompt from training slides and asks a colleague to check every draft (Beginner). The other has documented their own workflow, trains colleagues and tracks improved response rates (Power User). The difference is not in self-assessment but in verifiable patterns of behaviour over several months.

Ask for specific evidence: documents, screenshots, prompts, metrics, short recordings of AI workflows.
Use a simple grid where employees self-rate and managers rate, then discuss differences.
Document "Case A vs. Case B" examples for each level so rating differences between managers become smaller.
Connect ratings to existing processes in talent management and development conversations.
Store ratings and examples centrally — not in local spreadsheets — so calibration and audits are possible.

Example role profiles and target levels

Not every role needs to reach Champion. Realistic target levels per skill domain keep training focused and prevent overambitious expectations that breed frustration. The table below shows starting points — adjust for your context and risk appetite.

Role	AI fundamentals	Prompting & workflows	Data & privacy	Tool proficiency	Collaboration & change	Governance & risk
HR generalist	Practitioner	Practitioner	Practitioner	Practitioner	Beginner	Beginner
People manager (any function)	Practitioner	Practitioner	Beginner	Practitioner	Power User	Practitioner
Knowledge worker IC (e.g. Marketing, Finance)	Practitioner	Practitioner	Beginner	Power User	Beginner	Beginner
Leadership team / C-level	Champion	Power User	Champion	Practitioner	Champion	Champion

A data-heavy role in Finance or Legal typically needs higher target levels in "Data & privacy" than a comparable role in Internal Communications. Always adapt the table to your specific business context.

Growth signals and warning signs

Clear signals help you see who is ready for the next level and where risk sits. Promotions and greater AI responsibilities should follow visible patterns of behaviour over time, not one impressive demo.

Hypothetical example: an HR business partner automates their monthly people analytics report, shares the template, supports two other BPs in setting it up and documents GDPR-safe data flows. Over two quarters this pattern shows clear Power User behaviour — a solid signal for greater responsibility.

Growth signals: repeatedly delivers AI-supported outcomes with fewer errors and less supervision.
Multiplier effect: colleagues use their prompts, guides or templates without needing them every time.
Scope expansion: moves from team-only use cases to cross-function or cross-country workflows.
Compliance maturity: proactively involves the data protection officer and works council rather than asking at the last minute.
Warning signs: inconsistent quality, ignored policies, hidden experiments or resistance to feedback.

DACH rollout: BetrVG, GDPR and EU AI Act Article 4

Rolling out an AI skills matrix in DACH means balancing speed with co-determination, GDPR and psychological safety. Three legal layers are relevant:

EU AI Act Article 4 — AI literacy obligation (effective 2 February 2025): Providers and deployers of AI systems must take measures to ensure their staff have sufficient AI literacy (Article 4 EU AI Act). The obligation is a "best efforts" requirement — what matters is demonstrating that measures were taken. Four factors must be considered: technical knowledge and experience, education and training background, the specific context of use, and the groups affected by the AI systems. As the Haufe analysis notes, this means for HR in practice: a needs assessment, role-specific training, a governance framework and documentation. Violations can be subject to fines of up to €15 million or 3% of global annual turnover from 2 August 2025 onward.

BetrVG — co-determination rights: When using the matrix beyond pure development purposes, several sections of the Works Constitution Act apply. Section 94 BetrVG requires works council approval for personnel questionnaires and general assessment guidelines — relevant when skills are captured systematically with rating tools. Section 87(1)(6) BetrVG applies if ratings are linked to performance tracking or behavioural monitoring. Sections 96–98 BetrVG give the works council co-determination rights over any derived training measures. Clarify early whether the matrix is used for development only or also for formal evaluation: this determines which consultation steps are required.

GDPR — data minimisation and retention: Individual skill ratings are personal data. Define together with the data protection officer which evidence can be stored, where and for how long. For training design and benchmarking, aggregated and anonymised reports are sufficient. Individual ratings belong in performance processes — not in public sharepoints or local files.

Start small: one pilot department, a simple self-assessment and a workshop to align expectations.
Assign a clear owner (e.g. L&D or HRBP) and schedule an annual review to update the matrix.
Use anonymous, aggregated reports for training design; keep individual ratings within performance processes.
Connect results to AI training programmes — practical frameworks are described in the guide on AI training for employees.
Integrate the matrix into talent development processes rather than running it as a separate tool.

Interview questions by skill domain

You can also use the AI skills matrix template for hiring or internal moves. Behaviour-based questions reveal whether candidates have real experience or have only watched demos. For a People Manager role you might ask: "Tell me about a situation where you helped your team adopt a new AI-based way of working. What did you do specifically — and what changed?" Strong answers mention specific workflows, doubts that were addressed, metrics and lessons learned.

AI fundamentals: "Describe a situation where you learned a new AI tool or concept to solve a concrete work problem. What did you do, and what happened?"
Prompting & workflows: "Give an example where you iterated your prompts until the output was reliable. How did you approach that?"
Data & privacy: "Tell me about an AI use case where GDPR or data protection questions came up. How did you handle it?"
Tool proficiency: "How do you use tools like ChatGPT or Copilot in your working week? What has changed as a result?"
Collaboration & change: "Describe a moment when you involved colleagues in testing or improving an AI workflow. How did that go?"
Governance & risk: "Share a situation where you stopped or changed an AI idea because you saw risks. What triggered that?"

Probe consistently: "What was the result?", "How often do you do this?" and "What would you do differently today?" Candidates at higher levels talk about patterns, stakeholders and metrics — not just individual tasks.

Team check-ins and calibration sessions

To keep the AI skills matrix template alive rather than letting it become a one-off spreadsheet, you need light, recurring check-ins. They reduce rating differences between managers and build shared standards over time.

A practical format: once per quarter, all People Managers in a business unit meet for 60 minutes. Each manager brings two real AI use-case examples with evidence, proposes levels, and the group discusses and aligns. HR notes patterns and training needs.

Run short, recurring calibration sessions by function (HR, Sales, Operations) using real examples.
Use behavioural anchors from the matrix rather than vague labels like "senior" or "high potential".
Run basic bias checks: compare ratings across gender, age, location and manager.
Document final levels and key evidence in your performance or talent management system.
Turn shared gaps into targeted AI training paths — for example the curriculum described in the guide on AI training for HR teams.

Conclusion

A well-designed AI skills matrix template creates clarity: moving from "be more digital" to "this is what level X in domain Y looks like — with concrete behavioural anchors and evidence". Employees understand which behaviours lead to the next level. Managers gain a fairer basis for feedback, performance reviews and promotions. At the same time, the organisation systematically fulfils its AI literacy obligations under Article 4 of the EU AI Act — including the documentation that makes compliance demonstrable.

To get started: pick a pilot team in the next four to six weeks, run self- and manager assessments and hold a first calibration session using real use cases. Over three to six months you can derive focused AI learning paths and start reflecting promotion decisions against the matrix. Within six to twelve months the matrix can be rolled out to more business units, integrated into performance cycles and updated together with the data protection officer and works council in a first annual review.

FAQ

How do we start using the AI skills matrix template without overcomplicating things?

Start with one team and the six domains only. Ask everyone for a quick self-rating and one to two concrete examples per domain. Managers add their view. Discuss differences in 1:1s and then run a short calibration session to align levels and collect training needs. Keep documentation light at first and add detail once people understand the language and see the value.

What does Article 4 of the EU AI Act require from companies?

Article 4 EU AI Act (effective 2 February 2025) requires providers and deployers of AI systems to ensure their staff have sufficient AI literacy — proportionate to their technical knowledge, education, the context of use and the groups affected. It is a "best efforts" obligation: demonstrating measures taken (needs assessment, training, governance, documentation) matters most. Violations can be subject to fines of up to €15 million or 3% of global annual turnover from 2 August 2025 (Article 4 EU AI Act).

Which works council rights apply when introducing an AI skills matrix in Germany?

When capturing competencies systematically with rating tools, Section 94 BetrVG requires works council approval for personnel questionnaires and general assessment guidelines. If ratings are linked to performance monitoring, Section 87(1)(6) BetrVG applies. For derived training measures, the works council has co-determination rights under Sections 96–98 BetrVG. Clarify early whether the matrix serves development only or formal evaluation — that determines which consultation steps are required.

Should AI skill ratings influence promotions and salary decisions?

Not at the very beginning. Use the matrix first as a development and training tool so employees feel safe being honest. Once behavioural anchors and evidence standards are well understood, you can gradually link specific levels in specific domains to promotion criteria for relevant roles. Always combine AI skills with overall performance and impact — never as a standalone score.

How do we avoid bias when assessing AI skills?

Bias shrinks when decisions rely on observable behaviours and shared rubrics. Use the matrix as a checklist: "Which examples prove this level?" Collect evidence from multiple sources where possible, not only manager opinions. Run calibration sessions across teams and monitor rating patterns by gender, age, location and manager. Clusters suggest missing shared standards — not necessarily individual errors.

How often should we reassess AI skills and update the matrix?

Reassess key roles at least once per year, ideally linked to your regular review cycle. For high-change teams (e.g. data, digital, HR), a light mid-year check can help. Review the matrix itself at least annually: sharpen behavioural anchors, add new tools and reflect regulatory changes (e.g. the EU AI Act high-risk obligations coming into full force on 2 August 2026). Treat the document as living: stable enough for consistency, flexible enough to adapt to new AI realities.

Jürgen Ulbrich

CEO & Co-Founder of Sprad

Jürgen Ulbrich has more than a decade of experience in developing and leading high-performing teams and companies. As an expert in employee referral programs as well as feedback and performance processes, Jürgen has helped over 100 organizations optimize their talent acquisition and development strategies.