Training tia
📄 Adding Data Sources

📄 Adding Data Sources

Data Sources provide the foundational knowledge that trains tia to deliver intelligent insights to your users.

tia has two types of data sources:

  • Immediate Answers: Real-time answers auto-generated from your documentation. tia leverages advanced AI technology to provide instant, contextually relevant responses drawn directly from your organization's knowledge base. By ingesting and training on a wide array of documentation sources - including technical manuals, API guides, and internal wikis - tia can rapidly synthesize information to address user queries on the spot. This capability ensures that users receive accurate, up-to-date information precisely when they need it, significantly reducing the time and effort required to find solutions within your platform.
  • In-Depth Learning: Recommend learning after your users get answers. Beyond offering immediate solutions, tia takes a proactive approach to user education by seamlessly integrating with your academy resources. After providing quick answers, tia analyzes the context of the user's inquiry and their interaction history to suggest relevant courses, learning paths, or in-depth resources from your educational content. This feature transforms each user interaction into an opportunity for deeper engagement and continuous learning, encouraging users to expand their knowledge and proficiency with your product or service over time.

You'll need to add at least one Immediate Answers data source to power the real-time question answering aspects of tia. In-Depth Learning sources are optional, but recommended.

To add a Data Source, from your Dashboard click Data Sources > Add Source on either the "Immediate Answers" or "In-Depth Learning" lists.

ℹ️ Training doesn't stop after tia indexes your content. Be sure to visit Improving tia for additional techniques on how to improve the answers provided by tia.

Adding content from a public URL

After clicking "Add Source", if you see the platform your URL-based content is hosted on, go ahead and click it. Otherwise, click "Other". Some data sources require additional info, but most just require a URL.

For the "Other" source, you have two additional options:

  • Sub-branch of your website: Choosing this means we'll only pick pages that contain the URL you provide, e.g. we won't crawl /about if you provide /help. This is the most common way to import only a subset of information, such as only your support documentation living at /support or blog posts living at /blog. Try not to include anything after the initial path, e.g. don't include /help/index, just /help will work best.
  • Entire website: Choosing this means we'll pick all pages on your website starting from the URL you provide. Do not include anything after /, for example: https://support.yourcompany.com/.

Once you click "Import" we'll immediately begin the import process, starting the 'crawl' from the URL you chose and following links on the page to discover other pages. You should see progress live in about ~60 seconds, but you can also click the provided button to receive an email when the import completes. Feel free to navigate away from the page at this point and we'll continue importing in the background.

ℹ️ One common 'gotcha' here is entering the wrong URL, leading to a small number of documents being found. If your URL looks like yourcompany.com/help/get-started you'll want to remove the /get-started to ensure we pick up all help content. The final URL would then look like this: yourcompany.com/help

Here's a few tips for success while adding data sources:

  • Our importer is designed to import product information pages for online courses/events and support articles. It is not designed to import content from generic sources such as news sites, Wikipedia, etc.
  • Only input URLs you are authorized to access, such as your own support site. Never add a data source you do not have the rights to.
  • We will stop importing after 1000 crawled URLs or if it takes longer than an hour to crawl your site.
  • If we don't find any pages, or fewer than expected, double check the URL you provided and ensure it is inclusive of all the content you are looking to include in tia. Removing suffixes like "/main", etc. can help tia discover more content.

Adding content from Notion

ℹ️ We recommend using Notion as an Immediate Answers source. If used as In-Depth Learning, a user will need to log into Notion and have the proper access permissions to view the learning content.

⚠️ A Notion account can only be connected to a single tia account. Attempting to connect the same Notion account more than once will lead to the error message "The account selected is already connected to a different tia instance."

After clicking "Add Source", click "Notion" to begin the content selection process. You'll be prompted to log into Notion if you're not already logged in. From there, you'll see tia only has access to the pages you select. This allows you to share only specific content with tia, keeping any unselected content private and inaccessible to tia. On the next page, you can select the specific pages tia will train on. From there, the import will begin. You should see progress live in a few seconds, but you can also click the provided button to receive an email when the import completes. Feel free to navigate away from the page at this point and we'll continue importing in the background.

Adding content from Google Drive

ℹ️ We recommend using Google Drive as an Immediate Answers source. If used as In-Depth Learning, a user will need to log into Google Drive and have the proper access permissions to view the learning content.

⚠️ A Google account can only be connected to a single tia account. Attempting to connect the same Google account more than once will lead to the error message "The account selected is already connected to a different tia instance."

After clicking "Add Source", click "Google Drive" to select content from Google Drive. You'll be prompted to accept some permissions. Note tia will only be able to see the specific Google Drive files you select. Any unselected content will be private and inaccessible to tia.

Here's a breakdown of the permissions requested:

  • See, edit, create, and delete only the specific Google Drive files you use with this app: tia will use this permission to train on the file(s) you select on the next step. Note tia only views the files you share for the purposes of training and never modifies or deletes them
  • Associate you with your personal info on Google: This permission allows tia to verify a user's identity and receive a unique identifier for the user.
  • See your primary Google Account email address: This permission allows tia to display which account is connected as a Data Source when you return to the page.

After accepting the permissions, you'll select which file(s) tia has access to and we'll begin the import process. Note our Google Drive integration is optimized for documents, PDFs, and presentations. It is not designed for spreadsheets or other file types at this time.

Adding content from a PDF

PDF is a good format to import data into Tia if the content is not available on the web or is not available to the public. Clicking "Uploaded PDF" will present you with a field to drag and drop a PDF, or you can click to browse for your file. Just click 'Import' from there and we'll begin the import process. This will only take a minute or two for smaller PDFs.

ℹ️ We accept PDFs up to 20mb. PDFs can be up to 20 pages long for free plans and 1000 pages for paid plans. If you would like to upload a different file format, try our Google Drive integration.

Adding content from Guru

ℹ️ We recommend using Guru as an Immediate Answers source. If used as In-Depth Learning, a user will need to log into Guru and have the proper access permissions to view the content.

After clicking "Add Source", click "Guru" and you'll be prompted to enter your Guru Email Address and User Token. We use this information to load the collections available in your Guru account. To generate a User Token, go to your Guru account and click on Manage > Apps and Integrations > API Access > Generate user token.

After entering your Guru Email Address and User Token you'll be able to select which collection(s) you would like to add to tia. You can also choose to include only verified cards. Simply select the collections and we'll begin the import process.

Note there can be some overlap between Guru collections and tia, for example you could add Zendesk to tia and as a collection in Guru. We would recommend avoiding any overlap and only selecting one primary source for each data source.

Importing a private data source

If your data source is private, we would recommend converting it to PDF and using the "Upload a PDF" option. Most systems can convert any source of data to a PDF via their "Print" functionality, which also has the ability to save the printout as a PDF.