GDPR and AI in customer service: what you may do with data
Legal basis, data processing agreements, EU storage and the AI Act's transparency rules for chatbots. What must be in place before AI reads your tickets.
Contact centres and citizen services get the same questions every day: waste collection, building permits, water meters, invoices. Supportifier gathers the questions as tickets in one inbox. AI answers the resident from what you have reviewed, and a member of staff takes over the rest.
Supportifier is customer service software for contact centres and citizen services in municipalities and municipal companies. Email, chat, forms and phone become tickets in one shared inbox. AI answers residents only from published, reviewed information and hands over to a member of staff when the material is missing. Customer data is stored within the EU/EEA, a data processing agreement is signed and tickets are never used for AI training. The price follows ticket volume, not the number of staff.
Most of what reaches a municipal contact centre has an answer that is already on the website or in a routine. The same goes for customer service in a municipal company for water, waste, energy or housing. When the answer is reviewed, the resident can get it straight away.
The principle is the same as in the rest of Supportifier: the AI answers only from what you have reviewed yourselves, otherwise a member of staff takes over.
The information below describes the service as it is today and is written so it can be pasted into your documentation. What we lack is stated here too.
All storage is within the EU/EEA. To create an answer, the ticket text is sent to Anthropic, a US company. The text is not stored there after the answer has been created and is not used for training. Include that in your own assessment of the third-country question; documentation is provided on request.
The price follows ticket volume, not the number of staff, so the cost can be estimated before you ask us. Calculate the contract value over the whole contract period, with options, and check it against your threshold for direct award.
It is a ticketing system for customer service and contact centres: questions from residents by email, chat, form and phone become tickets with an owner, status and priority. It is not a records register or an operational system and does not replace them. Official cases are still handled in your own systems.
The chat answers only from articles you have reviewed and published, with a link to the source. If it finds no material it shows your own message and offers a way to a member of staff. Answers are labelled Generated by AI, so the resident knows who they are talking to.
All customer data in the service is stored within the EU/EEA. To create an answer, the ticket text is sent to the language model at Anthropic, a US company. The text is not stored there after the answer has been created and is not used for training. Include that in your own assessment; the list of sub-processors is provided on request.
Not today. Sign-in uses a Google account, restricted to the email domains you have approved. SAML, OIDC and SCIM against your own identity solution are not available. If that is a mandatory requirement for you, Supportifier does not fit yet, and we would rather say so now than in a tender.
The resident-facing surfaces, that is the chat, the form, the help centre and the customer portal, have not been audited against WCAG. We therefore cannot certify that they meet the Swedish act on accessibility of digital public services. The website supportifier.se has its own accessibility statement, but it does not cover the service.
You decide that from the contract value. The price follows ticket volume and the public plans are on the pricing page, so the value can be estimated in advance. Payment by invoice is quoted in the Enterprise setup. Ask for a written quote for your volume and contract period, and compare it with your threshold.
You can choose automatic deletion of tickets, and you can search for and delete a person's tickets. At the end of the contract you can request an export of tickets and the knowledge base. Whether a ticket is a public document and how long it must be kept is for you to decide under your records plan; Supportifier does not make that assessment.
Tell us which question takes time today. We go through what a first step could look like.
See how the knowledge work, the review and the first channel fit together.
How the rollout works