Observr Data Processing Agreement
Last updated: 11 November 2026
You run Indx yourself, and Bionic AS ("Indx", "we") has no access to your installation. Observr is the one exception. When you turn it on, your server sends us summaries of your search for analysis, and those can contain personal data. We then process that data on your behalf: you are the controller and we are your processor. This agreement sets out how, as Article 28 of the GDPR requires.
The agreement applies however you have Observr: through a Pro subscription, where it is part of the Pro subscription terms, through Enterprise, or through access we have granted on a free licence. It takes effect when you turn Observr on, and covers nothing else: the Indx software itself involves no processing by us.
1. What we process, and why
- Observr reads the search statistics of the datasets you choose and proposes changes to your search: synonyms, field settings and Search or ask settings. That is the only purpose.
- What your server sends: what visitors searched for, as typed, with counts; record names (the field that names a record, up to 120 characters); field names and roles, and the eight commonest values of a field whose values repeat (up to 80 characters each); your own settings and text; and what your users type in the Agent panel.
- What it never sends: whole records, and who a visitor is. No session, visitor id, IP address or email of a visitor.
- Whose data: visitors to your search, people named in record names or field values (an author, for example), and your users of the Indx console. A visitor can type anything into a search box, a name or an email included.
- Observr is not meant for special categories of personal data. Do not turn it on for a dataset whose record names or common field values are such data.
- The code that builds what your server sends is open source, in
IndxServer/Services/Observr, so you can see exactly what leaves.
2. Your instructions
- We process the data only to provide Observr to you, on your documented instructions: this agreement, and your choices in Observr (which datasets are on, and what you ask). If we think an instruction breaks data protection law, we tell you at once.
- We do not use the data for our own purposes, and it is not used to train any model.
3. Confidentiality and security
- Everyone at Indx with access to the data is bound to confidentiality by their employment contract, also after it ends. Access is given only to those who need it to run Observr.
- All traffic is encrypted over HTTPS, and your server signs in to Observr with its licence token. Data is kept apart per licence and stored encrypted in Microsoft Azure. Access to Azure requires two-factor sign-in, our machines have encrypted disks, and every change to Observr is reviewed by both developers.
4. Helping you
- We help you answer people who use their rights under the GDPR, such as access and erasure. A request that reaches us directly is passed on to you.
- We help you with security, breach notification, data protection impact assessments and prior consultation, as far as the nature of the processing allows (Articles 32 to 36).
5. Personal data breaches
We tell you without undue delay, and within 48 hours of becoming aware of a breach that affects your data, by email to the address on your licence account. We say what happened, which data and how many people it concerns, the likely consequences and what we have done, and send what we do not yet know when we know it.
6. Sub-processors
You authorise the sub-processors below.
- Microsoft (Azure): runs Observr and stores its asks, in the Norway East region.
- OpenRouter, Inc.: passes each ask to the language model, through its EU service, which processes and stores it in the EU only. Covered by the EU Standard Contractual Clauses and OpenRouter's EU-only processing addendum.
- Anthropic (Claude), through OpenRouter: the language model that writes the answer, processed in the EU.
We tell you at least 30 days before we add or replace a sub-processor. If you object and we cannot meet the objection, you can turn Observr off and end this agreement at no cost. We hold every sub-processor to the same obligations as this agreement, and answer to you for them.
7. Where the data is
In the EU and EEA. We do not transfer it out without your written consent. OpenRouter and Anthropic are based in the United States; their contracts include the EU Standard Contractual Clauses and a commitment to process the data in the EU only.
8. Audits
We give you the information you need to see that this agreement is kept. You can audit us, yourself or through an independent auditor you choose, with reasonable notice and at your own cost. We answer in writing first, and an audit on site takes place when written answers are not enough.
9. How long we keep it
- The text of each ask and its answer are deleted after 90 days. After that we keep only numbers with no personal data: when, for which licence, what kind of ask and what it cost.
- Ask us and we delete your data sooner. When this agreement ends, we delete it within 30 days, backups included, and confirm it in writing.
- What Observr found is stored in your own server's database, and is yours to keep or delete.
10. Start and end
The agreement applies from when you turn Observr on, and for as long as we process data for you through it. Turn Observr off under Admin > Observr and your server sends nothing more; what we hold is then deleted as in section 9.
11. General
- Where this agreement and the terms you use Indx under differ on personal data, this agreement wins. Liability follows those terms, within what the GDPR allows.
- If you want a signed copy, write to us and we will sign one with you.
- Norwegian law applies, and Oslo District Court is the venue if we cannot agree.
Processor: Bionic AS, org. nr 979 925 166, Von der Lippes gate 6, 0454 Oslo, Norway