Back to Blog

Customer Data in AI Tools: What Is Allowed and How Approvals Protect You

Henri Jung, Co-founder at Superkind
Henri Jung

Co-founder at Superkind

Superkind logo on an abstract orange haze texture

Customer data in AI tools is the question every team lead, DPO and managing director faces the moment someone on the team pastes a customer email into ChatGPT or forwards a support ticket to an AI assistant. What is allowed? What needs a contract? And what happens when the AI sends a reply to a customer without anyone checking it first? This article explains the rules - GDPR Article 6, data processing agreements, the EU AI Act - and shows how Superkind’s ready-made AI team for customer data handles this work with built-in approvals and European hosting.

Key takeaways

Get a ready-made AI team for customer data - Superkind comes with an AI team (Ticket Sorter, CRM Keeper, Follow-Up Writer, Privacy Scout) that works with your customer data automatically and asks you before anything sensitive goes out. Start for free with unlimited users, or build your own.

Keep control, keep data in Europe - hosted in Europe, GDPR compliant, aligned with the EU AI Act, no training on your data. You decide what needs approval - complaints and deletions wait, routine work runs on its own.

Legal basis first - every use of customer data in AI needs a legal basis under GDPR Article 6, plus a data processing agreement (DPA) under Article 28 with the AI provider1.

No training on your data - choose AI providers that contractually guarantee they do not use your customer data for model training. Free-tier tools often do.

Approval is the control point - an AI employee that waits before sending a customer reply, merging a CRM record or deleting data gives you the same oversight a human handoff does.

Your AI team for customer data

#customer-service
Ticket SorterAI employee
New ticket in Zendesk
Monday morning: 34 new tickets sorted. 21 standard replies sent (delivery status, password reset). 8 routed to sales (upsell signals). 5 complaints flagged for your review.
Auto-replied21
Routed to sales8
Waiting for you5 complaints
Runs automatically - complaints and refunds wait for your OK
Example: Ticket Sorter at work in Slack

What is customer data in AI tools?

Customer data in AI tools means personal data from your customers - email addresses, names, order histories, support messages, CRM records, call notes - that an AI system processes to do work for your team. Under GDPR, any processing of this data requires a legal basis and, when a third-party provider is involved, a data processing agreement1. The strongest form of AI for this work is an AI employee that does the task in your CRM, your helpdesk and your inbox, with an approval step before anything sensitive goes out. One example: Superkind, an AI employee you write to in Outlook, Teams, Slack or the web app - with 1,700+ integrations into Pipedrive, HubSpot and Zendesk, hosted in Europe, GDPR compliant.

How it works with Superkind: your AI team for customer data

You write to Superkind by email, in Teams, in Slack or in the web app, and Superkind does the task in your programs - with 1,700+ integrations, including Pipedrive, HubSpot, Zendesk and Freshdesk. It works with your CRM instead of replacing it, hosted in Europe and GDPR compliant.

With Superkind you don't start from zero: it comes with ready-made AI employees for customer data work. They run automatically in the background, and you decide where they ask you first. Use them right away, adapt them or build your own.

Ticket Sorter

AI employee for support ticket triage

Support tickets pile up over the weekend: refunds, shipping questions, complaints, feature requests. The Ticket Sorter reads every new Zendesk ticket, checks the customer history in HubSpot, assigns a priority and routes it to the right person. Replies to standard questions go out on their own. Complaints and refunds wait for your OK.

Works inSlackHubSpot+ 1,700 more
#customer-service
Ticket SorterAI employee
New ticket in Zendesk
Monday morning: 34 new tickets sorted. 21 standard replies sent (delivery status, password reset). 8 routed to sales (upsell signals). 5 complaints flagged for your review.
Auto-replied21
Routed to sales8
Waiting for you5 complaints
Runs automatically - complaints and refunds wait for your OK
Example: Ticket Sorter at work in Slack

CRM Keeper

AI employee for CRM data maintenance

Half the contacts in your CRM have outdated phone numbers or missing company names. Every morning the CRM Keeper checks new and changed records in Pipedrive, fills in gaps from email signatures and public sources, and flags duplicates. Merges and deletions wait for your approval.

Works inMicrosoft TeamsOutlook+ 1,700 more
CRM UpdatesSales
CRM KeeperAI employee
Routine - every morning at 7:00
Morning CRM check complete. 12 contacts updated with current job titles, 3 duplicates found (Meier/Meyer and Lang GmbH), 1 bounce address removed.
Contacts updated12
Duplicates found3
Bounced removed1
Merge 3 duplicates?Meier/Meyer (same email), Lang GmbH (2 entries)ApproveReject
Example: CRM Keeper at work in Microsoft Teams

Follow-Up Writer

AI employee for sales follow-ups

The sales call ends, and the follow-up sits in your head until Friday. The Follow-Up Writer picks up every new call note in Pipedrive, drafts a follow-up email with the agreed next steps, and sends it from your inbox. Quotes and discount offers wait for your sign-off.

Works inOutlookGmail+ 1,700 more
Follow-up: Fischer AG demo
Follow-Up WriterAI employee
New call note in Pipedrive
Follow-up for Fischer AG drafted and sent. Summary of demo points included, next step: trial access by Thursday. Deal stage in Pipedrive moved to "Proposal".
Sent toj.fischer@fischer-ag.de
Deal stageProposal
Done - standard follow-up, sent without approval
Example: Follow-Up Writer at work in Outlook

Privacy Scout

AI employee for data compliance checks

A customer asks what data you hold on them, or the DPO wants a weekly report on where personal data flows. The Privacy Scout runs through your tools, finds where customer data is stored, and prepares a report. Deletion requests and subject access requests wait for your review before anything is changed.

Works inMicrosoft TeamsHubSpotExcel+ 1,700 more
Privacy ReportsOperations
Privacy ScoutAI employee
Routine - every Monday at 9:00
Weekly data flow report ready. Customer PII found in 4 systems: HubSpot (1,842 contacts), Zendesk (926 tickets), Pipedrive (714 deals), Excel export on SharePoint (outdated, last touched 90 days ago).
Systems with PII4
Outdated exports1 (SharePoint)
Open deletion requests2
Delete 2 records?Deletion requests from K. Brandt and M. YilmazApproveReject
Example: Privacy Scout at work in Microsoft Teams

Every employee is ready to use right away. Or build your own on the platform.

Safe by default

Customer emails, CRM contacts and support tickets are personal data. Superkind processes them in Europe, with a DPA through the Trust Center, no model training on your data, and approval before anything sensitive goes out.

How the rules work: customer data and AI

  1. Legal basis (Art. 6 GDPR) - before handing customer data to any AI tool, you need a legal basis. The most common: contract performance (Art. 6(1)(b)) when the AI supports an existing customer relationship, or legitimate interest (Art. 6(1)(f)) for tasks like ticket triage and CRM maintenance. Consent (Art. 6(1)(a)) is an option when neither applies1.
  2. Data processing agreement (Art. 28 GDPR) - a DPA is mandatory when a provider processes personal data on your behalf. It defines the purpose, duration, type of data, security measures and subprocessor rules. Without a DPA, forwarding a customer email to an AI tool is a GDPR violation2.
  3. No training on your data - many free-tier AI tools use input data for model training by default. Under GDPR, using customer data to train a model without a separate legal basis is problematic. Check the provider’s terms and opt out or choose a provider that never trains on your data6.
  4. Transparency (Art. 13/14 GDPR) - your customers have a right to know that their data is processed by an AI system. Update your privacy policy and, where the AI communicates directly with customers, make the AI use clear1.
  5. AI literacy (EU AI Act Art. 4) - since February 2025, companies that deploy AI systems must ensure their staff has sufficient AI literacy to use the tools responsibly. This applies to everyone who works with customer data in an AI tool4.
  6. Approval as a control point - an approval step before the AI sends a reply, merges a record or deletes data gives you the same oversight as a human handoff. It is not legally required for every task, but it is the simplest way to stay in control of sensitive actions.

The practical test

Ask three questions before handing customer data to an AI tool: (1) Which legal basis under Art. 6 applies? (2) Is there a signed DPA with the provider? (3) Does the provider train on your data? If any answer is missing, stop until it is sorted.

AI tools for customer data compared

ApproachWhat it doesHosting / Data
AI employee (Superkind)Does the task in your CRM, helpdesk and inbox; approval where you want it; 1,700+ integrationsEU, GDPR, no training, DPA
General chatbot (ChatGPT, Claude)Answers questions, drafts text; does not act in your systemsCheck with vendor
CRM-native AI (HubSpot Breeze)Drafts inside the CRM; limited to one systemProvider’s own terms
Automation (Zapier, Make)Moves data between tools by rules; no reasoning or approvalCheck with vendor

An AI employee like Superkind works across all your customer data systems. A chatbot helps you draft, but does not act. A CRM-native tool is locked to one platform.

Examples: customer data and AI in practice

  • Customer emails - a customer writes about a late delivery. Superkind reads the email, checks the order status in Pipedrive and the tracking in DHL, and drafts a reply with the current status. The reply waits for your approval before it goes out. Legal basis: contract performance (Art. 6(1)(b)).
  • CRM contacts - new leads come in from a trade fair. Superkind enters them in HubSpot, checks for duplicates and fills in missing fields from email signatures. Duplicate merges wait for your OK. Legal basis: legitimate interest (Art. 6(1)(f)), backed by the DPA with Superkind.
  • Support tickets - a Zendesk ticket about a billing error. Superkind reads the ticket, finds the invoice in the system, and prepares a correction. The correction waits for approval. Legal basis: contract performance, plus the DPA.
  • Deletion requests - a customer asks you to delete their data (Art. 17 GDPR). Superkind finds records across HubSpot, Zendesk and Pipedrive, prepares the deletion and waits for your sign-off. The Privacy Scout makes sure nothing is missed.

What it costs and what you need

  • DPA with every AI provider - check whether your current AI tools have a signed DPA. If not, get one or switch. Superkind provides a DPA through its Trust Center at trust.superkind.ai.
  • Legal basis documented - for each AI use case with customer data, write down which Article 6 basis applies and why. This takes an afternoon with your DPO, not a project.
  • AI literacy training - since February 2025, the EU AI Act requires that staff using AI have sufficient knowledge. Brief your team on what customer data the AI sees, what it does with it, and where the approval step sits4.
  • Superkind pricing - start for free with up to 150 hours of work, no credit card needed. Business plans from EUR 50 per month for 50 hours of work, unlimited users. Enterprise plans with SSO, audit log and custom DPA on request (as of November 2026). See /pricing for details.

Data protection and EU hosting

Superkind is a sovereign European choice. Hosted and processed in Europe under European law, GDPR compliant, aligned with the EU AI Act (Regulation (EU) 2024/1689), no training on your data. AI models from OpenAI, Anthropic, Mistral and Google, all processed in Europe. No lock-in to one model vendor. Everything you build on the platform - your Company Brain, routines and apps - belongs to you. For details and the DPA, see trust.superkind.ai.

How to start

Conclusion

Customer data in AI tools is not a grey area. The rules are clear: a legal basis under GDPR Article 6, a data processing agreement under Article 28, no training on your data, and transparency towards your customers. The approval step is what makes the difference between handing work to an AI employee and hoping nothing goes wrong.

Superkind gives your team an AI employee that works in Pipedrive, HubSpot, Zendesk and 1,700+ other systems, hosted in Europe, with a DPA from day one and approval where you set it. Start with the first customer task today - for free, with unlimited users, no credit card.

Frequently asked questions

You need a legal basis under Article 6 GDPR. The most common ones are contract performance (Art. 6(1)(b)) when the AI supports an existing customer relationship, and legitimate interest (Art. 6(1)(f)) for tasks like ticket triage or CRM maintenance. A data processing agreement (DPA/AVV) with the AI provider under Article 28 is mandatory. Superkind provides a DPA through its Trust Center and does not train on your data.

Yes. Superkind is hosted and processed in Europe, GDPR compliant, aligned with the EU AI Act (Regulation (EU) 2024/1689) and does not train models on your data. AI models from OpenAI, Anthropic, Mistral and Google are processed in Europe. A DPA is available through the Trust Center at trust.superkind.ai. With 1,700+ integrations into tools like Pipedrive, HubSpot and Zendesk, your customer data stays inside a European setup.

A DPA (in German: Auftragsverarbeitungsvertrag, AVV) is a contract required by Article 28 GDPR whenever a provider processes personal data on your behalf. It defines what data is processed, for what purpose, by whom, and what security measures apply. Without a DPA, handing customer emails, CRM contacts or support tickets to an AI provider is a GDPR violation.

Only with a valid legal basis and your explicit agreement. Many free-tier AI tools use input data for model training by default. Under GDPR, using customer data for training without consent or a separate legal basis is problematic. Superkind does not train models on your data, and that is contractually guaranteed in the DPA.

The approval step lets you decide which actions with customer data need your OK before they happen. For example: a reply to a customer complaint, a CRM record merge or a deletion request. Routine tasks like ticket sorting or follow-up emails for existing deals can run on their own. You set the rules, Superkind follows them.

Superkind starts at EUR 0 with up to 150 hours of work free, no credit card needed. Business plans start from EUR 50 per month for 50 hours of work, with unlimited users. Enterprise plans come with dedicated support, SSO and a custom DPA. See superkind.ai/pricing for current plans (as of November 2026).

Related articles

Henri Jung, Co-founder at Superkind
Henri Jung

Co-founder of Superkind, the AI employee teams write to in Outlook, Teams and Slack. Henri helps companies, from small teams to the Mittelstand, hand their routine work to AI - without new tools, consultants or an in-house AI team.

Send Superkind your first customer task.

Start for free in Outlook, Teams or Slack. Or book a 30-minute demo to see how it works with your CRM and your customer data.