✦ In short
Teammeter uses AI where it saves the most time: building the skill catalogue, defining roles and filling in skill profiles. The AI proposes, people decide.
- Skill catalogue: descriptions, translations, level descriptions, skill relationships and their transfer scores
- Roles: generated from a role name or a job description
- Profiles and trainings: skills extracted from a CV, smart skill suggestions for each profile, skills suggested for a training
- Control: you review AI proposals, choose the AI provider and can switch AI off
Skill management lives from good data: a clear skill catalogue, well-defined roles and up-to-date skill profiles. Building this foundation is often the part that takes the longest. Writing a description for hundreds of skills, translating them, defining what level 1 to 5 means for each one, or entering the skills of every new employee by hand: this is where many projects lose momentum.
This is exactly where we use AI in Teammeter. Not to evaluate people, but to take the repetitive work off your hands. Here is an overview of the AI features available today.
Building the skill catalogue faster
✨ Autocomplete for skills
When you create or edit a skill, the AI suggests a description, translations into the languages of your platform, synonyms, a category and a skill type. You review the suggestions in the form and save only what fits. Each type of suggestion can be switched on or off in your company settings.
📶 Descriptions for each proficiency level
What does “level 3” in Kubernetes mean compared to “level 4”? Clear level descriptions make self-assessments and manager assessments comparable. The AI drafts one description per level of your scale, and you adjust them in the skill editor before saving.
🔗 Relationships between skills
Someone who knows Java can learn C# quickly. The AI proposes such relationships between the skills of your catalogue, for example similarity or specialisation. Only skill names and descriptions are used, no personal data. You can edit or remove any relationship at any time.
For each relationship, the AI also estimates two transfer scores between 0 and 1, one for each direction. A transfer score answers one question: if someone masters skill A, how quickly can they deliver real work in skill B?
The two directions are usually different. Mastering Java gives you a head start in Kotlin, but knowing Kotlin teaches you a little less about Java. In the same way, a broad skill does not imply every specialisation, while a specialisation tells you a lot about the broader skill. Skill matching uses these scores to give partial credit for related skills, so a person who knows Vue.js still appears for a role that asks for React, with a lower match than an exact fit. Like the relationships themselves, every transfer score can be reviewed and adjusted in the skill editor.
Defining roles in minutes
🏷️ Generate a role from its name
Enter a role name such as “Cloud Architect” and the AI proposes a description and the skills the role requires. Teammeter shows which of these skills already exist in your catalogue and which ones are new, so your catalogue stays consistent.
📋 Create a role from a job description
Many companies already have job descriptions. Paste one into Teammeter and the AI turns it into a structured role proposal with name, description, skills, tasks and responsibilities. Nothing is saved until you have reviewed it.
Filling skill profiles and trainings
📄 Import skills from a CV
New employees start with an empty profile. With the CV import, they upload their CV and the AI extracts skills and certificates from it. The text is read in the browser and pseudonymised before it is processed, and the CV itself is not stored. Nothing is added to the profile until the employee has confirmed, edited or deleted each proposal.
🎯 Smart skill suggestions
Teammeter recommends skills for each profile by analysing people with similar skill profiles in your company. If many people who know Kubernetes also know Terraform, Terraform is suggested to someone whose profile lists Kubernetes but not yet Terraform. The suggestions are calculated statistically from your own company’s skill profiles, without sending any data to an AI provider. Each person decides which suggestions to add to their profile.
🎓 Skill suggestions for trainings
A training only helps skill planning if you know which skills it develops. Based on the name and description of a training, the AI suggests the matching skills with a target level. You review, edit or delete the suggestions before saving.
Insights for teams
After a team health check, the AI summarises the results and suggests actions for the team. The answers are collected anonymously and are not linked to individual people, so the summary is about how the team works, not about who answered what.
🛡️ How you stay in control
- The AI proposes, you decide: AI results are suggestions that you review before they are used.
- Your choice of AI provider: Teammeter works with AWS Bedrock, OpenAI, Azure OpenAI and OpenAI-compatible providers such as Mistral AI, as well as self-hosted models.
- AI can be switched off: administrators can deactivate AI features for their company.
- Little personal data: most AI features only work with catalogue data such as skill, role and training descriptions. The CV import is pseudonymised before processing.
- Quality checked every day: automated tests check the quality of the AI results daily.
If you introduce AI tools in your company, your teams also need the right skills to use them. Read more in our articles on AI literacy under the EU AI Act and AI skills in the company.
❓ Frequently asked questions about AI in Teammeter
No. The AI supports the creation of catalogue data and proposes skills from a CV, which the employee confirms. Matching scores between people and roles are calculated transparently by a fixed formula, without AI. The AI only proposes the transfer scores between skills that this formula uses, and you can review and adjust them.
Yes. AI features can be deactivated for your company, and all data can also be entered by hand.
That depends on your configuration. Teammeter supports AWS Bedrock, OpenAI, Azure OpenAI and OpenAI-compatible providers such as Mistral AI, as well as self-hosted models. We are happy to discuss which option fits your requirements.