How the AI risk calculator works
Start with your job title, then describe a normal working day. Artha asks short follow-up questions about your responsibilities, industry, distinctive skills, computer use and how much you work with other people. The conversation is designed to gather useful context rather than judge your career from a title alone. Usually, the assistant aims to collect enough detail over five to seven exchanges.
Your report includes an indicative score, reasons some tasks may be exposed to automation, reasons human involvement still matters, skills to explore and alternative roles to research. You may also receive an estimated outlook or timeline. These are AI-generated suggestions, not a prediction of what your employer will do. Read the explanations before focusing on the number.
Our risk assessment methodology
The tool sends your conversation to Claude with a structured prompt about work in India. The prompt asks the model to consider repetitive versus creative tasks, judgement, human interaction, physical work, sector and location. Its scoring bands are Low (0–30), Medium (31–60), High (61–80) and Very High (81–100). These bands are instructions to the model, not independently validated statistical thresholds.
There is no published probability model behind the displayed percentage. It should be understood as a directional assessment, not your measured chance of losing a job. The tool does not consult your employer's staffing plans or verify current local vacancies. Different descriptions, or another run of the conversation, can produce different answers. No occupation is guaranteed to be unaffected by technology.
Turn the report into a small next step
Make a list of tasks you could practise with an approved AI tool and tasks where you provide checking, accountability or relationships. Choose one useful skill and test it on a small project before buying a course. Discuss workplace changes with a manager or trusted colleague when appropriate. Do not resign, take on debt or make a major career switch because of this score alone.
Illustrative example: two accounting roles
Imagine an accountant who spends most of the day copying invoices into spreadsheets. The report might flag repetitive information handling and suggest learning reconciliation or reviewing automated outputs. Another accountant who advises clients and handles unusual compliance questions may receive a different assessment despite sharing the same title.
This is a fictional illustration of how task descriptions influence the conversation, not a real user's result or an occupational benchmark. Neither example establishes a risk percentage. The useful takeaway is to describe the parts of your work that require context, responsibility and judgement as carefully as the routine parts.
AI Risk Score: your questions answered
Is the risk score a probability of redundancy?
No. It is a model-generated indicator based on your answers and our prompt. A score of 70 does not establish a 70% chance of losing your job.
Can people with the same job title receive different scores?
Yes. Responsibilities, industry, location and human involvement can differ. Give concrete examples of your work rather than relying only on a job title.
How long does the conversation take?
The prompt usually aims for five to seven exchanges. Your answers and service availability affect the time; a completed report within a fixed number of minutes is not guaranteed.
Does a low score mean my job is AI-proof?
No. A lower score reflects this conversation's assessment, not permanent safety. Tasks, business priorities and technologies can all change.
What should I do if a result feels wrong?
Check whether your answers described your actual responsibilities. Try again with more detail, compare the reasoning with workplace evidence and ask a qualified adviser before making major decisions.
What information should I avoid sharing?
Do not include employer secrets, customer data, identity documents or sensitive personal details. Conversations are processed by an AI service and the application may retain chat logs.