Trust
Private by architecture.
Security and People & Talent teams should get straight answers. This page covers how The Impression Advantage Coach protects the people who use it, the data it holds, and the AI that powers it.
Request our security overviewFour principles.
Confidential by design
What someone shares with their coach is theirs.
Enforced, not promised
Privacy rules live in the platform's database itself, so no screen, report or administrator account in the platform can open one person's coaching.
Insight without exposure
Organizations see patterns only across groups of 10 or more, never individuals.
For development, never evaluation
The Coach helps people navigate trust windows. It does not score, rank or report on them.
Three pillars
Privacy, security and responsible AI.
Privacy
- What an employee shares with their coach is visible only to them. Their manager, their People & Talent team and their employer cannot see it.
- People & Talent teams see patterns across groups of 10 or more people, never individuals. Smaller groups show nothing, so no one can be singled out.
- Each organization's data is kept separate from every other organization's.
- People can ask for a copy of their own data, or for it to be deleted, by emailing privacy@impressionadvantage.com.
- Our own team does not access coaching content.
Security
- Data lives in a private database with no public internet access.
- Data is encrypted in transit and at rest.
- Coaching conversations are encrypted again, with their own keys, before they are stored, and every use of those keys is recorded in an audit log.
- The application reaches the database with short-lived credentials, not long-lived access keys.
- Every change to the platform is tested automatically, including tests that prove these privacy rules hold.
- Data is stored in the United States on Amazon Web Services.
Responsible AI
- The coach works within a defined framework, the eight trust windows, with clear boundaries.
- It does not handle job searches, legal, medical or harassment matters, and points people toward the right help.
- It is built for development, never evaluation: individual coaching content is never reported to the employer.
- Coaching conversations are not used to train AI models. Our AI provider, Anthropic, is contractually barred from training on them.
Before pilots launch
The platform completes an independent security review and penetration test, and begins a SOC 2 Type I audit.
If someone is in crisis
If someone tells the coach they are thinking about suicide or self-harm, the coach right away gives them the 988 Suicide & Crisis Lifeline, and recommends professionals contact their local emergency number if they are outside the US, and stays with them in plain, caring words instead of continuing to coach. A separate check reads every message for these signs, so the referral does not depend on the coach alone. We count how often this happens, with no name or company attached, so we can make sure it works. Employers never see it.
Certifications are listed here as they are earned. Security documentation is available on request.
Frequently asked questions
Can our organization see what an employee discusses with the coach?
No. Coaching is private to the employee. Managers, People & Talent teams and administrators cannot open an individual's coaching; the platform's database enforces this, not only our policy.
Is the manager's conversation private too?
Yes. What a manager discusses with the companion is private to them. Their direct report, People team and administrators cannot read it.
What do People & Talent teams actually see?
Patterns across groups of 10 or more people in a trust window, shown in an analytics dashboard. Groups smaller than that show nothing, so no individual can be identified.
Where is our data stored?
In the United States, on Amazon Web Services, in a private database with no public internet access. Data is encrypted in transit and at rest. AI processing also runs only in the United States: to write each reply, our AI provider, Anthropic, processes the conversation there and deletes it within 30 days, except where it must keep it longer to enforce its usage policy or to meet the law.
Is coaching data used to train AI?
No. We do not train AI models on coaching conversations, and Anthropic's commercial terms bar it from training its models on them.
How is AI used, and what won't it do?
The coach uses AI to guide people through their trust window, within the trust windows framework. It does not evaluate or score people, report individual content to the employer, or handle job searches, legal, medical or harassment matters.
How do we get your security documentation?
Request our security overview and we will walk your team through it.