AI at Worknano: what it is for, where it runs, and its limits
Worknano uses AI to help teams work with their own business data. This page says plainly what that AI is for, what it must not be used for, which models it uses and where they run, and how people stay in charge. It is our statement of the intended purpose of Worknano's AI features.
What Worknano's AI is for
- Assistant: answers questions about your records, files and goals, and drafts and summarises text.
- Agents: carry out tasks your team sets up, such as follow-ups, summaries and reports, with the tools and data an admin gives them.
- Field help: suggests values and text while you fill in a form.
- Search by meaning: finds documents and records that match what you mean, not only the exact words.
- Voice and images: turns speech into text and creates images on request.
- Builders: turn a description into a module, a form or a website you can then edit.
What it must not be used for
Worknano's AI is not made for decisions about people. Do not use it to:
- select, rank or reject job applicants;
- rate the performance or behaviour of individual employees, or decide on promotion, pay, termination or the allocation of work based on that;
- decide on access to credit, insurance, education or public services;
- identify people from their face, voice or other body features, or infer their emotions.
Tracking team and company goals is fine. Rating or ranking individual people is not. Under the EU AI Act these uses count as high-risk. If a customer uses Worknano's AI for one of them, the customer takes on the duties that come with that use.
Models and where they run
- You choose the model family for your workspace, and each agent can have its own model.
- EU families keep AI processing in the EU.
- Other families can send prompts to model providers outside the EU, for example in the USA, with the safeguards the GDPR requires for international transfers. They include models from OpenAI, Anthropic, xAI and DeepSeek, and open models such as Llama and Qwen, hosted in Germany or the USA.
- Your own model: you can also connect a model of your choice, including one your company runs itself.
- Search by meaning uses an embedding model in the EU, and its index stays in your workspace database.
- Voice input and image creation use OpenAI models.
Your data
- We do not use your data to train AI models.
- A model only receives what it needs to answer the request at hand.
- Customers get the full list of our subprocessors with their data processing agreement.
People stay in charge
- Assistants and agents only see and change what the signed-in user is allowed to.
- Changes the assistant suggests are shown as previews, and a person confirms them.
- Admins decide which agents exist, what they may reach and which model they use.
- Each workspace has a monthly AI usage limit.
Limits
AI can be wrong, and it can sound sure when it is. Check important results before you rely on them, and do not use AI output as the only basis for legal, medical or financial decisions.
The EU AI Act
We build Worknano's AI features on general-purpose models from other companies. Under the EU AI Act that makes us the provider of these features, and your company the deployer when your team uses them. Within the intended purpose above, Worknano's AI is not a high-risk system. We train our own team in the use of AI, and we tell people when they are dealing with AI where that is not already obvious.
This page describes Worknano. It is not legal advice for your own use of AI.