Clear about what we do — and what we don’t do yet.
Our assistants talk to your customers, so we hold ourselves to plain, verifiable statements. Here is where we stand today.
CURRENT STATUS
Developing our security and privacy framework.
We don’t claim certifications or compliance we haven’t achieved. We will publish each control as it is in place.
Our AI assistant handles the routine work. You stay in control.
It works inside the boundaries you set. When a request needs judgment, it escalates rather than guessing or making a promise you didn’t authorize.
On its own, it…
- Answers common questions from your approved information
- Explains services, published prices, business hours and policies
- Collects contact details and requirements, with notice
- Organizes inquiries and routes them to the right place
- Recognizes when a request is beyond it and directs the customer to a person
It never…
- Invents prices, availability, policies or commitments
- Says a booking, cancellation or update succeeded unless your system confirms it
- Goes beyond the permissions and rules you set
- Diagnoses conditions, prescribes treatment or replaces a professional’s judgment
- Discloses customer information or uses it for anything you haven’t authorized
- Carries on when a request needs human judgment
Developing our security and privacy framework. No certifications claimed.
Security & TrustWhat customers are told in every chat.
| Topic | What customers are told |
|---|---|
| AI disclosure | They are talking to an AI assistant, not a person. |
| What’s collected | Name, contact details, messages and appointment details. |
| Why | To answer questions, handle inquiries and support bookings or follow-ups. |
| Who sees it | The business they contacted and authorized service providers. |
| Sensitive information | Not to share medical, financial or other sensitive details in chat. |
| Retention | [Retention period or criteria] Content neededRetention period or criteriaOwner: Legal counsel |
| Their rights | [How to request access, correction or deletion] Content neededHow to request access, correction or deletionOwner: Legal counsel |
| Talking to a person | How to ask for a person when the assistant can’t help. |
Where data sits and who can reach it.
Hosting and storage
[Hosting provider and data storage location.]
Access
[Who can access customer data (US and India teams), and the safeguards in place.]
Records
[What each conversation and action writes to the record, how long it is kept, and who can read it.]
Our principles.
These principles describe how we build and run our assistants. Where a safeguard is still being put in place, we say so.
Transparency
Customers always know when they are talking to AI.
Human oversight
Assistants handle routine work; sensitive or consequential decisions involve a person.
Privacy and security
Collect only what’s needed and protect it with appropriate safeguards.
Accuracy
Answer from approved information, and escalate when unsure.
Accountability
Clear owners for monitoring performance and handling concerns.
Continuous improvement
Review and improve as technology, needs and regulation change.
What security reviewers ask us.
Straight answers to the questions practices and procurement teams ask first.
Is Cognelith HIPAA compliant?
Not yet. HIPAA readiness has not been independently evaluated. Please don’t use the assistant to collect protected health information. We’ll assess HIPAA requirements before offering healthcare-specific products.
Do you have SOC 2 or ISO 27001?
No. We have not completed a SOC 2 examination or started ISO 27001 certification.
Do you take card payments?
Not in Capista at launch. If payments are added later, a third-party payment processor will handle card details.
Where is our data stored, and who can access it?
[Answer to come from the CTO — see “Your data” above.]
Need documentation for a review?
Practices and procurement teams can ask for what we have so far.