Claude is Anthropic's AI assistant. You give it a question, a document or a task, and it writes, analyses and thinks along with you. We work with it ourselves and are very happy with it, especially for long texts, analysis, code and documents. It also writes well in Dutch. Where it falls short is speech: dictating in Dutch does not work well, but there are workarounds. Using Claude with a team? Choose a business plan and have someone check what comes out.
The details on models and plans were checked on 8 October 2026. They change often, so see the sources at the end for the current position.
What is Claude in three sentences?
Claude is a family of AI models from Anthropic, together with the apps and tools around them. You use it through the Claude app on web, desktop and mobile, through Claude Code if you build software, and through an API when a developer wants to build it into their own software. There are also extensions for Chrome and Microsoft 365 (Anthropic).
Anthropic offers several models in different sizes. On 8 October 2026 these are Fable 5.1 for the most demanding reasoning and agent work, Opus 5.5 for long-running knowledge work and coding, Sonnet 5.5 for a balance of speed and quality, and Haiku 5.5 for high-volume small tasks such as sorting and extracting data (Anthropic, models overview). Model names change every few weeks, so do not tie yourself to one name.
What do users say about Claude?
We work with Claude ourselves and are very happy with it. That is our experience, not independent proof: we have an interest in AI. Look at what is happening around us this year and you can see that Claude is being used seriously, and also that checking is still needed:
- Large organisations are rolling it out. In May 2026 KPMG announced an alliance with Anthropic and is giving more than 276,000 employees worldwide access to Claude, including for clients in tax and private equity (KPMG). That is KPMG's own announcement, but it shows that even an organisation where security and oversight weigh heavily is working with it.
- The market is paying attention. When Anthropic released a new version in February 2026 that it says is also suited to legal work, shares in legal software companies fell, Wolters Kluwer by almost 20%. A researcher at TU Delft told the Dutch legal outlet Mr. Online that it will not go that fast: the question is who makes knowledge available in a reliable and verifiable way (Mr. Online, in Dutch).
- The best score is not a guarantee. In June 2026 Aithos, a Dutch AI research non-profit, tested twelve AI models acting as agents on scenarios under the EU AI Act and the GDPR. Claude Opus 4.7 scored best of the twelve, with 54% compliance. That is the best result, and still only just over half (Euronews).
This does not prove that Claude is the best choice for your business. It shows that it is no longer a toy, and that even the best models need someone to check them.
What is Claude good at in a business?
According to Anthropic, Claude reads Word documents, PDFs, spreadsheets, presentations, images and code, and creates editable files such as documents and slides. It can also pull in information from connected tools such as Gmail, Google Drive, Slack, Salesforce and Microsoft Office (Anthropic). A single conversation can hold roughly 555,000 words, according to Anthropic. That is enough for an entire case file.
That makes it useful for tasks such as:
- Long texts. Summarising a report or contract, comparing two versions, rewriting a text in your house tone.
- Analysis. Working through a spreadsheet, flagging anomalies and explaining what stands out.
- Code. Writing a small script, tracking down an error, explaining existing code.
- Creating documents. A first draft of a quote, memo or presentation, which someone then reviews and finishes.
Suppose an account manager receives five quote requests. Claude can set each request against the price list and prepare a draft reply. The account manager checks the figures and sends it. The work gets faster, but responsibility stays with a person. This is a fictional example, not a measurement.
Does Claude work well in Dutch?
In our experience Claude works well in Dutch for text: you write in Dutch and Claude answers in Dutch. Anthropic recommends naming the language you want, and asking for "idiomatic speech as if it were a native speaker" to get more fluent sentences (Anthropic, multilingual support). Anthropic does not publish a score for Dutch, so test with your own texts: is the tone right, does Claude use your industry terms, and does it choose between formal and informal address the way you would?
Speech is another matter. The Claude app has no Dutch interface: Anthropic lists eleven languages and Dutch is not among them. Voice mode has its own language setting, and Dutch is not listed there either (Anthropic Help Center, August 2026). Anthropic still describes voice support for languages other than English as a beta (Anthropic Help Center). Users report that dictating in Dutch returns English gibberish (LinkedIn post by a Dutch user). That is awkward if you would rather talk than type.
Fortunately there are workarounds:
- Dictate outside Claude. The dictation feature on your phone or computer turns your Dutch speech into text, which you paste into Claude or leave in place.
- Use a dictation app. In the replies to that LinkedIn post, users mention apps such as WhisperFlow, which works in any app, Claude included. Note that your voice then passes through an extra supplier, so check the same data arrangements.
- Speak English. If that suits you, voice mode works in English as normal.
- Just type it. For short requests it costs little extra time.
This is the position on 8 October 2026. Voice in other languages is in beta and may change quickly.
Claude for teams: what do you need to arrange?
A personal account and a team account are not the same thing. The difference lies in administration and in what happens to your data.
Claude Team is meant for 2 to 150 users. It costs $20 per user per month on annual billing or $25 monthly, and according to Anthropic it includes central billing, single sign-on, connector management and, by default, no training on your data. Audit logs, SCIM and custom retention periods only come with Enterprise (Anthropic, plans & pricing).
For commercial products, Anthropic says it does not use your data to train its models, unless you join the Development Partner Program. The customer is then the controller of the data and Anthropic the processor (Anthropic Privacy Center, March 2026). On the personal plans, a setting decides whether your conversations are used to improve Claude (Anthropic Privacy Center).
"Not used for training" is not the same as compliance. It says nothing about a data processing agreement, retention period or location. So agree three things before the team starts:
- Account. A business plan with an administrator, and no personal accounts for work.
- Data. Which client and company data may go in, and which never may. Put that in a short AI policy.
- Review. Who reads what Claude produces before it goes out.
Also allow for usage limits. Claude works with a five-hour session limit and a weekly limit. Long conversations, attachments, searches and multi-step tasks use more of it (Anthropic Help Center). Does your team have little experience with AI assistants yet? The Getting started with AI workshop helps you on your way.
What is Claude not, or what can it not do?
- It is not a source of truth. Anthropic says Claude can sometimes give incorrect or misleading answers, including quotes that look authoritative but have no basis. Users should not rely on it as a "singular source of truth" and should scrutinise high-stakes advice (Anthropic Help Center). For answers based on a web search, check the sources it cites.
- It does not know your company by itself. Claude only knows what you give it or connect to it. An assistant that answers staff questions from your own documents has to be set up for that, for example in an AI workplace. For what this can look like in Teams, see An AI chatbot that knows your company knowledge.
- It does not take over responsibility. Whoever sends a quote, text or analysis remains responsible for it. Put an owner on every use.
Claude alongside ChatGPT and Copilot: how do you choose?
None of the three wins everywhere, and the ranking shifts with every new release. So make a team choice on your own tasks, not on a comparison table. Pick three recurring tasks, have five colleagues do them in each tool, and judge on accuracy, usability and administration. For how that works and what the plans cost, see ChatGPT, Claude or Copilot for your team: how to choose.
Want to use Claude safely in your team? Book a conversation.
Sources
- Anthropic, Claude product overview, accessed 8 October 2026.
- Anthropic, Plans & pricing, accessed 8 October 2026.
- Anthropic, Models overview, accessed 8 October 2026.
- Anthropic, Multilingual support, accessed 8 October 2026.
- Anthropic Help Center, How to use Claude in your preferred language, 6 August 2026.
- Anthropic Help Center, Using voice mode, accessed 8 October 2026.
- Anthropic Help Center, Usage limit best practices, accessed 8 October 2026.
- Anthropic Help Center, Claude is providing incorrect or misleading responses, 16 March 2026.
- Anthropic Privacy Center, Does Anthropic act as a data processor or controller?, 16 March 2026.
- Anthropic Privacy Center, Is my data used for model training?, 16 March 2026.
- KPMG, KPMG and Anthropic sign global alliance and launch Digital Gateway Powered by Claude, 19 May 2026.
- Mr. Online, Schokgolf in legal tech door nieuwste Claude-editie?, 12 February 2026 (in Dutch).
- Euronews, AI agents actively ignore EU law to achieve goals, study finds, 2 June 2026.
- LinkedIn, post by a Dutch user about Claude and Dutch speech, accessed 8 October 2026.
