Answers you can trust, boundaries you control
edrak is built to be a reliable, managed assistant — clear about what it knows, where it stops, and when a human steps in.
This page explains how edrak handles knowledge, boundaries, and data. It is not a legal policy — a formal privacy policy is reviewed separately by legal counsel before publication.
How an answer is handled
Every reply is grounded in your approved knowledge — and escalates to your team whenever it should.
Principles we operate by
A few commitments that shape how edrak is built and run.
Knowledge-based answers
edrak answers from your company knowledge base — not random guesses.
Boundaries you define
You decide what the assistant answers and what gets handed to a human.
Your data stays yours
Your business data is not used to train public AI models.
Managed setup
Setup, hosting, knowledge-base prep, channel connection, and launch — handled for you.
Visible human handoff
Escalation is built into the experience, following the rules you set.
No exaggerated claims
No promises of perfect answers, full team replacement, or specialized legal, medical, or financial advice.
How we handle your data
Practical safeguards applied across setup, hosting, and day-to-day operation.
Encrypted in transit and at rest
Conversations and stored data are encrypted in transit over TLS and at rest on disk, so information is protected on the wire and in storage.
Least-privilege access
Access to systems and customer data is limited to authorised personnel on a need-to-know basis, with activity logged.
Managed, monitored hosting
We run on reputable cloud infrastructure with monitoring, patching, and backups handled for you as part of the managed service.
Clear data retention
Data is kept only as long as needed to run your assistant. You can request an export or deletion when you offboard.
No training on your data
Your business and conversation data are never used to train public or shared AI models — only to run and improve your own assistant.
Vetted subprocessors
We work only with established providers — hosting, AI model, and analytics — under confidentiality and data-processing terms.
What edrak does
- Answers strictly from your approved knowledge base.
- Stays within the topics and boundaries you configure.
- Hands conversations to your team based on your escalation rules.
- Keeps a clear record of how answers are produced.
- Lets customers know when a human will follow up.
What edrak never does
- Invents facts or answers outside your knowledge base.
- Sells or shares your data with advertisers.
- Trains public AI models on your conversations.
- Claims to replace your whole team or give specialised legal, medical, or financial advice.
- Hides that it is an assistant where disclosure is expected.
Your data, your control
Reach out any time to exercise these — we'll verify your account first.
Access
Ask what data we hold about your account.
Export
Request a copy of your knowledge base and configuration.
Correction
Update inaccurate information whenever you need to.
Deletion
Request deletion of your data when you offboard.
Security questions, answered
The questions teams ask most before they connect a channel.
Formal policies
This page explains our approach. For the binding details, read our policies.
Have specific security or data requirements?
Tell us what you need and we'll walk through how edrak handles it.
Talk to us about security