Our AI agents propose a risk class for every AI feature and collect the evidence.
Example data
Under the EU AI Act, AI intended to filter applications or evaluate candidates is usually high-risk AI. The same applies to AI intended to monitor the performance and behavior of workers. From 2 December 2027, providers need quality management, technical documentation and a conformity assessment for it.
Our AI agents prepare this with ISO 42001 on one control set. A human approves every decision.
AI that infers emotions at the workplace has been prohibited since 2 February 2025. Exceptions exist only for medical or safety reasons. Fines up to €35 million or 7% of worldwide annual turnover, whichever is higher.
Check transparency duties, for example for chatbots.
As the deployer, you use the system according to the instructions for use, assign trained oversight staff, keep logs and inform workers in advance. From 2 December 2027.
Risk management, quality management system, technical documentation, conformity assessment, CE marking and registration. Conformity assessment is internal control, without a notified body. From 2 December 2027. Fines up to €15 million or 3% of worldwide annual turnover, whichever is higher.
Annex III AI that profiles natural persons is always high-risk. The exception for simple support tasks does not apply.
If your AI only performs a narrow procedural task or only prepares an assessment, an exception can take it out of the high-risk class. Document your reasons.
Whether your AI falls under Annex III depends on its intended purpose. This helper is not legal advice.
Tick what runs in a way you can prove. Your result stays in your browser.
The foundations are still missing
Book a demoClassified, evidence still missing
Book a demoReady for customers and audits
Book a demoThe Digital Omnibus moved the high-risk duties to 2 December 2027. Your customers plan their deployer duties now and need your documents for it.
AI that places targeted job ads, filters applications or evaluates candidates is listed in Annex III. So is AI that monitors performance or allocates tasks by behavior.
The narrow-task exception does not apply if your AI profiles natural persons.
AI that infers emotions at the workplace has been prohibited since 2 February 2025.
Deployers use the system according to your instructions and keep your logs. Employers inform workers and their representatives before use.
Our AI agents propose a class for every AI feature and map the provider duties. They collect the evidence from your tools.
GetTeddy drafts the AI register from your product and connected tools. Each team confirms what it builds and who owns it.
For each feature GetTeddy proposes a risk class and your role, with documented reasons. Legal confirms the result.
The Policy Agent drafts the AI policy and AI roles and links them to your existing policies. ISO 42001 and ISO 27001 run on one control set.
The Evidence Agent collects change history, model documentation and data sources from connected tools. Auditors see the lifecycle in practice.
The Questionnaire Agent answers security and AI questionnaires from your approved evidence. You check every answer before it goes out.
Product, data science and sales ask in their own words. GetTeddy answers from the AI register, documentation and controls.
| Typical | With GetTeddy | |
|---|---|---|
| Classification | Debate without reasons | Risk class with documented reasons |
| Documentation | Spread across wiki and tickets | Collected from connected tools |
| Customer questions | Every questionnaire by hand | Answered from current evidence |
| ISO 42001 | A separate project next to ISO 27001 | On the same control set |
No. Under Annex III point 4 of the EU AI Act, high-risk covers AI for recruitment and selection, and AI for decisions on promotion, termination or task allocation or for monitoring performance. Scheduling software that does not evaluate people is not covered. The duties for providers of high-risk AI apply from 2 December 2027.
Only if your AI does not profile natural persons. An exception in the EU AI Act can take a narrow procedural task, or a task that only prepares an assessment, out of the high-risk class. Annex III AI that profiles people is always high-risk, and ranking candidates is often profiling. Document your reasons either way.
The high-risk duties for recruiting AI apply from 2 December 2027. The Digital Omnibus moved this date. Emotion recognition at the workplace has been banned since February 2025. Transparency duties have applied since 2 August 2026. For generative AI already on the market before then, the marking duty applies from 2 December 2026.
No. As the provider of recruiting AI, you assess conformity yourself, through internal control and without a notified body. Before the market launch, you register the system in the EU database. These duties apply from 2 December 2027. GetTeddy prepares the technical documentation and the evidence for this internal control.
Your customers, as deployers, use the system according to your instructions, assign trained staff for human oversight and keep the logs for at least six months. Employers inform workers and their representatives before use at the workplace. These deployer duties apply from 2 December 2027.
Read next EU AI Act · ISO 42001 with GetTeddy · Compliance for AI companies · SOC 2 · AI in credit scoring for fintechs
We show you how it works on your own product.