The first responsible use decision is where not to use it.
Nothing automated reads the career file you submit for a CareerOS 360 package or signal audit. No parser, no scoring engine and no language model interprets that record or decides what it means. Coaching programmes differ, with your consent, as set out below.
Four judgements stay with a person.
The service is designed around an accountable reviewer because these decisions require context, challenge and a source trail.
What a claim means
A title or number can mean very different things across organisations. The reviewer asks what was owned, what changed and what source supports the answer.
What stays in
A claim is not retained because it sounds strong. It stays because the record can support it and the person is prepared to stand behind it.
What the target role asks
Role alignment is a judgement about the mandate in front of the reader. It is not a keyword match against an advert.
What is released
A person reads every draft before delivery and remains responsible for the wording that reaches you.
If the workflow changes, this page changes first.
A future tool cannot quietly become part of career-record interpretation. Any material change must be described in plain language before it is introduced, including what data it receives, why it is used, who reviews the output and how a person can object.
Current position: no parser, automated scoring engine or language model reads career records submitted for CareerOS 360 packages or the signal audit; coaching programmes are covered below. The browser-only self-check reads only the answers selected on that page, sends nothing and stores nothing; it deliberately returns a band rather than a Profile Signal Index.
How your record is handled
Illustrative photography. People shown are not CVPRO clients or coaches.
More on how we work
CareerOS agent builds
How the CareerOS agents use AI.
CareerOS 360 packages (P1–P3) and the signal audit use no language model. A CareerOS agent build, which you choose separately, does — and this section names every AI step in it.
What the agents do
- Read the CV and LinkedIn PDF you send, and draft the audit, role strategy and package files from them
- Research relevant public professional, role, company and opportunity information for your case only
- Run fourteen quality checks and report an honest status for every file
Where it runs, and who decides
- Provider: OpenAI, on the ChatGPT platform (model: GPT-6 Pro at the time of writing); processing may take place outside India
- We ask for your separate consent before an agent reads your material; you may decline
- A named CVPRO reviewer checks every claim and file before delivery. No automated verdict, ranking or hiring prediction is ever made about you
What the agents never do
- Invent claims, inflate titles or pick the larger of two conflicting numbers
- Infer protected traits or research private information, data brokers or personal social media
- Reuse your facts in another case, or claim to have sent emails or applications
The consent you give
Before an agent reads anything, your written scope asks you to agree to this, word for word:
“I consent to CVPRO processing my CV, LinkedIn PDF and related material with its CareerOS AI agents on OpenAI’s ChatGPT platform (processing may take place outside India) for this engagement only, including research of relevant public professional information. A named CVPRO reviewer checks every output; no automated decision is made about me. I can withdraw this consent at any time by writing to grievance@talproindia.com.”
Your choices
You can restrict research (for example, “no web research”), ask what was processed, or ask us to delete your case material by writing to the Grievance Officer at grievance@talproindia.com. The full standard is published at how the agents think.
One policy, three services.
CVPRO runs different services, and the honest answer to “does AI touch my material?” depends on which one you use. Here it is, service by service.
No parser, scoring engine or language model reads the record you submit. One named reviewer reads it, front to back.
Two in-house CareerOS agents read the CV and LinkedIn you send and draft the package on OpenAI’s ChatGPT platform. This happens only with your separate consent in the written scope; a named reviewer checks every claim and file before delivery, and no automated verdict is made about you. How the agents work.
AI tools may help prepare your evidence pack and session briefs from material you provide for that engagement. This happens only with your separate consent in the written scope; the tool and provider are named in your scope and on the Security and privacy page before first use; a person reviews everything before your coach relies on it. AI never coaches you and is never presented as a coach.
If this ever changes, this page changes first, before the new step is used on anyone’s material.