Skip to content
Insight AIInsight AI Docs

Quickstart: from data to answers

This walkthrough takes you from an empty workspace to asking questions about your own documents. It touches every core feature once; each step links to a deeper guide.

Before you start, have ready:

  • The connection details of an SMB network file share (server name, IP address or hostname, port, username, and password). Your IT team can provide these.
  • About 10 minutes.

  1. Create a group.

    Data sources belong to a group, so you need at least one. Go to Users, select Create Group, give it a name (no spaces, at least 2 characters) and a permission level, then select Create Group.

    → Details in groups & permissions.

  2. Connect a data source.

    Go to Data Sources → Create Data Source. Name the source, choose your group as the Owner Group, and fill in the SMB connection details (username, password, remote system name, IP address, port — 445 by default). Insight AI verifies the credentials immediately; if they’re wrong, the source isn’t created.

    Once created, open the source and check the Shares tab — you should see the folders and files on the share.

    → Details in connecting a data source.

  3. Create a collection.

    Go to Collections → Create Collection and name it (letters, numbers, and underscores only — for example company_handbook). This is the searchable index your documents will land in.

    → Details in collections.

  4. Create a pipeline.

    Go to Pipelines → Create Pipeline. Name it, add a description, and pick your Data Source and Collection. A pipeline connects exactly one source to one collection.

    → Details in pipelines.

  5. Run the ingestion.

    Open your new pipeline and switch to the Monitoring tab. Select Start Ingestion. You’ll see live progress: files discovered, files processed, and a progress bar. Supported documents (PDF, Word, PowerPoint, text, Markdown, CSV) are extracted, split up, and indexed for semantic search.

    When the run completes, the pipeline status turns active.

    The pipeline Monitoring tab after a completed run: progress 100%, 8 of 8 files processed

    → Details in running & monitoring ingestion.

  6. Add a model.

    If your organization hasn’t pulled a model yet, go to Models → Create Model and enter a model ID from the Ollama library (for example llama3.2). Watch the download progress; when it finishes, the model is benchmarked and available to everyone.

    → Details in managing models.

  7. Chat with your data.

    Select Chat in the navigation (opens a new tab). Use the model selector at the top to pick a model — if the list is empty, choose Add a model and add one of your organization’s models to your personal list. Then ask a question about your documents.

    Insight AI automatically searches your collections and gives the model the most relevant passages, so answers reflect your own content.

    A chat where the answer draws on details from an ingested policy document

    → Details in chatting with Insight AI.

Where to go next

Section titled “Where to go next”