AI Assistant for Business: Types, Uses and Data Protection

Blog · AI Assistant

An AI assistant for business is software that understands instructions in plain language and takes on office work: drafting emails, summarising meetings, searching documents and analysing spreadsheets. It comes in three forms: built into office suites, such as Microsoft Copilot or Gemini in Google Workspace; as a standalone chat such as ChatGPT; and as an assistant set up specifically for your business and its processes. Which one fits depends less on the technology than on three questions: where is your data processed, what does the assistant know about your business, and who signs off on its results?

In brief

There are three types: assistants built into office suites, standalone AI chats and assistants set up for your own business.
Under the GDPR, what counts is a data processing agreement, excluding your inputs from model training and where the data is processed.
Personal accounts are the wrong place for company data: German data protection authorities recommend company-provided accounts and devices.
Under Article 4 of the EU AI Act, any business using AI assistants must take measures to build AI literacy in its team.

Published 25 Sep 2026Nikolai Schöbel and Jeremias Burger9 min read

Nikolai Schöbel and Jeremias Burger

Co-founders of Scalableloops GmbH. Nikolai Schöbel leads online marketing and AI strategy, Jeremias Burger the AI architecture. Both build AI systems and train teams on them in their own agency work.

On this page

What is an AI assistant?
What types of AI assistants are there for businesses?
What is an AI assistant good for in the office?
How do you use an AI assistant in line with the GDPR?
Where does your data go with Copilot, Gemini and ChatGPT?
What does an AI assistant need to genuinely help your business?
When is an assistant tailored to your business worth it?
Does your team need training to use an AI assistant?
Frequently asked questions
How to find the right AI assistant
Where the information on this page comes from

Definition

What is an AI assistant?

An AI assistant is software built on a large language model that you talk to in everyday language and that takes work off your plate. You type something like “Summarise the email thread about this quote” and get a summary, a draft or an answer. The key word is assistant: it prepares the work, and the decision stays with you.

Two related terms often get mixed up. A chatbot on your website talks to customers, and since 2 August 2026 it comes with its own disclosure obligation, which we cover in our article on AI disclosure for chatbots. An AI agent goes a step further than an assistant: it completes several steps on its own and works with your software as it does so. The line between the two is blurred.

When this article talks about a personal AI assistant, it means the tool a person uses at their desk, not the customer chat on your website.

Overview

What types of AI assistants are there for businesses?

Most small and mid-sized businesses will come across three types. They differ in where the assistant works, what it knows about your business and where your data is processed.

Built into office suites. Microsoft Copilot sits inside Word, Excel, PowerPoint, Outlook and Teams, and Gemini is part of Google Workspace, including Gmail, Docs and Drive. The advantage: the assistant works where your documents already live and respects the access rights you have set there.

Standalone AI chats. ChatGPT and similar services run in the browser or as an app. They are ready in minutes, but they only know as much about your business as you tell them. For data protection, the deciding factor is whether your team uses a business account or personal accounts.

Assistants set up for your business. These are configured for one company, know its services, templates and processes, and work with the software already in place, from the mailbox to the inventory system. Getting started takes more effort, but the assistant fits the way you work rather than the other way round.

Feature	Built into office suite	Standalone AI chat	Assistant for your business
Examples	Microsoft Copilot, Gemini in Google Workspace	ChatGPT and similar chat services	An assistant configured for your company
Where it works	In Word, Outlook, Teams or Gmail, Docs, Drive	In the browser or its own app	In the software you already use
What it knows about you	Whatever is in emails, files and calendars the user can access	Only what you type or upload in the chat	Your services, templates, processes and rules
Access to other software	Mainly within its own office suite	Limited, usually copy and paste	The systems you choose to connect
Data protection	Business terms of the office provider, check the settings	For company data only with a business account and data processing agreement	Depending on setup, up to running on your own server
Getting started	Quick if you already use the office suite	Immediate	Needs a setup phase
Best for	Teams that do most of their work in the office suite	Individual tasks such as writing and research	Recurring processes that need company knowledge

Everyday use

What is an AI assistant good for in the office?

An AI assistant is strongest on tasks with a lot of text and a recurring pattern. For Copilot, Microsoft lists drafting, rewriting and summarising documents in Word, analysing data in Excel, summarising email threads in Outlook, and summarising meetings in Teams and capturing the agreed action items.

In practice that means the first draft of a reply to a customer, the notes from a meeting, finding a commitment buried in a long email thread, or analysing a spreadsheet nobody has had time for. The assistant mainly saves you the run-up; a person still adds the final touch.

The limits are just as clear: an AI assistant can be wrong and still sound convincing. In a representative Bitkom survey from November 2025, 42 per cent of people in Germany who use AI for search said they had received false or made-up information. For office work, that means anything that leaves the building, whether a figure, a commitment or a price, gets checked by someone on your team first.

AI has now reached most businesses. According to a Bitkom survey of 603 German companies with 20 or more employees, published in September 2026, 57 per cent use AI, up from 36 per cent a year earlier.

57 %

of companies in Germany with 20 or more employees use AI, up from 36 per cent a year earlier.

Bitkom, September 2026

Data protection

How do you use an AI assistant in line with the GDPR?

You can use an AI assistant in line with the GDPR if the contract, the settings and your internal rules are in order. In Germany, the framework comes from the Data Protection Conference (DSK), the body of the independent German data protection authorities, in its guidance “Artificial Intelligence and Data Protection” of 6 May 2024. It is aimed at organisations that use language models and chatbots.

If you use a cloud-based assistant, the provider processes data on your behalf, and the DSK says this will often count as processing on behalf of a controller. You then need a contract under Article 28(3) GDPR, and you may only work with providers that offer sufficient guarantees of appropriate technical and organisational measures (Article 28(1) GDPR).

The authorities also expect that a data protection impact assessment under Article 35 GDPR will often be needed when AI is used. For that you need information from the provider on how its system works, so make sure it is available when you choose a tool. The following points are drawn from the guidance:

01
01
Sign the contract
Put a data processing agreement in place with the provider before any personal data goes into the assistant.
02
02
Exclude training
Check whether your inputs and outputs are used for training and whether you can switch that off. The DSK considers tools that do not train on them preferable.
03
03
Use company accounts
Employers should provide accounts and devices for work use rather than leaving staff to use personal accounts.
04
04
Prefer closed environments
The DSK rates systems in which data stays in a contained environment, with access limited to a defined group of people, as preferable.
05
05
Write down the rules
Set out which assistants may be used for which purposes. Examples of permitted and prohibited uses make the rules easier to follow.

Providers

Where does your data go with Copilot, Gemini and ChatGPT?

The major providers draw a clear line between business and consumer offerings. This is the position as of 25 September 2026, based on their own documentation:

Microsoft Copilot: According to Microsoft, prompts, responses and the data Copilot accesses in Microsoft 365 are not used to train the underlying language models. Copilot only surfaces content the user has at least view permission for, and for EU customers it is an EU Data Boundary service. One exception concerns the Anthropic models Microsoft also offers: they are currently excluded from the EU Data Boundary and switched off by default for customers in the EU.

Gemini in Google Workspace: Google commits that your content is not reviewed by humans or used to train generative AI models outside your domain without your permission. Gemini follows your existing sharing settings: if a person is not allowed to see a file, the assistant cannot use it either.

ChatGPT: For its business plans, ChatGPT Business and Enterprise, OpenAI does not use your data for training by default. On personal accounts such as Free and Plus, training is switched on until you turn off “Improve the model for everyone” in the data controls. OpenAI offers a data processing agreement to business customers; there is none for personal ChatGPT accounts.

One point is often overlooked with built-in assistants: an assistant that respects access rights will find everything a person is allowed to see, including files that used to be hard to find. Microsoft explicitly advises keeping permissions, for example in SharePoint, set so the right people have access to the right content. Tidy up folder sharing before you switch on a built-in assistant for everyone.

Requirements

What does an AI assistant need to genuinely help your business?

An assistant is only as useful as what it knows about your business and what it is allowed to reach. A general-purpose chat knows neither your services nor your templates nor where things stand with your most important order. It produces generic text that someone then has to rework. Three things make the difference.

01
01
Knowledge of your business
Services, templates, contacts, tone of voice and the rules your business works by. The better this knowledge is maintained, the less your team has to fix.
02
02
Access to your software
Mailbox, calendar, file storage, inventory system or shop. Only with access can the assistant check the current status instead of guessing, and put results where your team needs them. It gets access only to what the task requires.
03
03
Human sign-off
The assistant prepares, a person decides. Anything that commits your business externally, such as quotes, promises and emails to customers, goes out only after review.

Our approach

When is an assistant tailored to your business worth it?

A tailored assistant pays off when you have a lot of recurring work that needs company knowledge: answering enquiries, preparing quotes, following up on cases, pulling documents together. If your team works almost entirely in office documents, the built-in assistant is the obvious place to start.

The three requirements above are the core of our AI agent system. It runs on your server, is independent of any single AI provider, knows your business and works with the software you already use. Nothing goes out without sign-off from a person on your team.

For the data protection question, that means your data stays in the environment you choose, and you decide which language model is used.

Obligations

Does your team need training to use an AI assistant?

Yes, to an appropriate extent. Since 2 February 2025, Article 4 of the EU AI Act has required businesses that use AI systems to look after their staff's AI literacy, regardless of sector or size and even if the team only uses an AI chat. Since 27 July 2026 an amended version applies: you must take measures that support the development of AI literacy, but you do not owe a specific level for each person. No certificate is required. The details are in our article on the AI literacy obligation under Article 4.

Separately, the German data protection authorities recommend raising staff awareness through training, guidelines and conversations about whether and how they may use AI tools. For an AI assistant, that means anyone using it should know which data may go in, where the assistant tends to get things wrong and who signs off on results.

In our AI training, teams practise exactly this on their own tasks, with the assistant they actually use.

Frequently asked questions

Which AI assistant is right for a mid-sized business?

It depends on where your team works. If most work happens in Microsoft or Google office tools, the built-in assistant is the natural starting point. For recurring processes that need a lot of company knowledge, an assistant set up for your business fits better. A standalone chat works for individual tasks such as writing and research, and for company data only with a business account.

Can I use ChatGPT with customer data?

Not with a personal account. For personal data you need a data processing agreement, which OpenAI only offers to business customers, and settings that exclude your inputs from training. According to OpenAI, training is off by default for ChatGPT Business and Enterprise. Also check with your data protection officer whether a data protection impact assessment is needed.

Is Microsoft Copilot GDPR compliant?

Microsoft treats Copilot as an EU Data Boundary service for EU customers and, according to its documentation, does not use prompts and responses to train its language models. Whether your use is compliant still depends on you: the contract, permissions, internal rules and, where needed, a data protection impact assessment. According to Microsoft, the additional Anthropic models are excluded from the EU Data Boundary and switched off by default in the EU.

What is the difference between an AI assistant and an AI agent?

An AI assistant works for you on request: you ask, it drafts. An AI agent completes several steps on its own and uses your software to do so, for example checking a case, creating a draft and filing it. The line is blurred. With both, a person should sign off on anything that goes out.

Can an AI assistant run on our own server?

Yes. Assistants set up for your business can be configured to run on your server and work with your existing software. The German data protection authorities consider closed systems, where data stays in a contained environment, preferable.

Do employees need training before using the assistant?

Article 4 of the EU AI Act requires businesses that use AI to take measures to build AI literacy. No certificate is required. A briefing on the specific assistant makes sense: which data may go in, where it makes mistakes and who signs off on results.

What now

How to find the right AI assistant

01
List the tasks
Write down the tasks in your office that come up every week and involve a lot of text: emails, meeting notes, quotes, reports.
02
Check what you have
Find out whether your office suite already includes an assistant and which personal AI accounts your team is already using.
03
Sort out data protection
Sign the data processing agreement, switch off training on your data and tidy up folder sharing.
04
Prepare your team
Write down what is allowed and train your team on the specific assistant. We can help with AI training or an AI agent system tailored to your business.

An AI assistant saves time where it knows your business and a person has the final say. Choosing the tool is the smaller part of the job; the bigger part is getting knowledge, access and sign-off right.