Skip to content
MAIAMAIADocumentation
EN
Open MAIA
MAIA / RESOURCES
View as Markdown

Start with AI, without the jargon

Understand what MAIA can do, try a first task and learn how to check an AI answer.

MAIA helps you work with information: summarize a document, prepare a draft, compare requirements or research a question. You describe a task in everyday language, provide the relevant context and review the result. You do not need to know how to program.

Start with a short document you are allowed to use, such as an internal meeting summary. Open a new chat, attach the document and ask:

Summarize this meeting for a colleague who was absent. Separate decisions, open questions and next steps. For each next step, include the owner and deadline only if the document states them. Otherwise write “not specified”.

Read the result beside the original. Check one decision, one date and one assigned owner. Then ask for a useful change: “Make this shorter and keep the unresolved questions.” The conversation lets you refine the same task.

Term Plain meaning What to do with it
Prompt The instruction you give MAIA State the task, context and desired result
Model The AI system that generates an answer Start with Auto; compare models when you have a reason
Context Information available for the current task Provide the relevant files, sources and instructions
Citation A link to material supporting an answer Open it and check that it supports the claim
Hallucination Plausible output that is wrong or invented Check facts rather than trusting fluent language
Connector A connection to an external service Connect your account and select the service for the task

A model produces text based on learned patterns and the context supplied to it. It is not a person who has independently checked every statement. MAIA's source tools can retrieve information; a generated answer still needs review.

Use files in chat for a particular conversation. Use a project when several related tasks need the same files and project context. Choose web search for public information that may have changed.

MAIA does not automatically have access to every company system. Integrations depend on workspace availability, your connected account and your permissions. Describe missing information explicitly instead of asking the model to guess it.

  • Does it answer the question you actually asked?
  • Are names, numbers, dates and quotations supported by the source?
  • Does it distinguish a fact from an assumption or a suggestion?
  • Is a qualified person needed to review the outcome before it is used?

Answer verification helps check source support. It is an additional check, not a guarantee that a decision is correct. Treat a draft as a draft until the responsible person has reviewed it.

Move from ad hoc conversations to repeatable prompts, Skills and workflows. Before sharing a reusable method, test it on several examples, including a case where the source is incomplete. The evaluation guide shows how to make that comparison fair.