We estimate work. We do not score your worth.
The free scan and paid report estimate which parts of a described work week may be delegatable, substantially accelerable, AI-assisted or primarily human with current general-purpose AI. The paid report is generated automatically and displayed immediately after verified payment. Neither product predicts dismissal, salary, promotion, job-loss timing or whether an employer will adopt AI.
What the estimate uses
Occupation baseline
A typical task mix informed by public occupational descriptions and task data, including the O*NET® database.
Your actual week
Your selected role, weekly hours and adjustments across eight work modes: research, writing, data, administration, meetings, judgment, relationships and physical/on-site work. The paid flow also uses your goal, tools, work archetype and the constraints you want to keep human.
Current AI capability assumptions
Different work modes receive different estimated delegation and acceleration factors. These are transparent product hypotheses used for an early transaction test, not universal scientific constants or a validated occupational standard.
Human review
The estimate assumes meaningful work still needs review, exception handling, accountability and context. It does not count raw model runtime as the full cost of work.
Automatic delivery does not make the estimate a fact
The instant report is a self-directed information product. It is not individually reviewed by a human before delivery and is not professional, employment or management advice. Its purpose is to identify plausible workflow experiments. A real benefit must be measured on a specific task.
Potential is not proven saving
“About 9–14 hours of this task mix appear AI-addressable.” This means the work looks suitable for delegation or acceleration, not that the time has already been saved.
Measured saving“The human baseline was 120 minutes. AI execution plus review and correction took 24 minutes, with agreed quality achieved in four of five trials.” We use this stronger language only after a real task trial.
The four output classes
Delegate
AI can plausibly execute a large share of a repeatable digital task, with defined review and exceptions.
Accelerate
AI can materially speed up preparation, analysis, drafting or quality control, while a person remains central.
Assist
AI can help with options, summaries, preparation or feedback, but is not a reliable owner of the outcome.
Keep human
Trust, accountability, physical execution, leadership, negotiation, taste or high-context judgment dominate.
Data source attribution
This product includes information informed by the O*NET 31.0 Database by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA), used under the CC BY 4.0 license. O*NET® is a trademark of USDOL/ETA. Česká zdravotní akciová společnost has modified and added to the information; USDOL/ETA has not approved, endorsed or tested these modifications.
Important limits
- Two people with the same title can have radically different work.
- AI capability varies by model, tool access, data quality, workflow design and error tolerance.
- Organizational permissions and system access can prevent implementation.
- High-stakes, regulated or confidential work may require additional controls or may be unsuitable.
- All results are informational and are not career, employment, legal or financial advice.