Question answering¶
With Alan AI’s Q&A service, you can quickly build an AI-powered Q&A agent that will leverage your data sources, both static and dynamic, to address user queries. The Q&A service employs a combination of question answering, semantic search and code generation by AI to automatically deliver relevant responses and maintain multi-turn contextual conversations.
The Q&A service can be helpful if you already have a pool of information that you want to use to:
Handle user requests
Offer 24/7 automated support
Onboard new customers and employees
Provide instructions, training and so on
The Q&A service is a low-code solution that enables you to build a Q&A AI agent with minimal coding. It only requires you to specify your data sources and a few parameters. Once configured, the Q&A service automatically crawls the data sources and ingests the information from them to create a private knowledge memory for your AI agent, which is then used to converse with users.
When building a Q&A AI agent, you can combine diverse types of data sources. The Q&A service accepts the following data formats:
Web pages
Plain text
PDF
JSON data
Alan Al provides several tools to build a Q&A AI agent:
Learn how to crawl static data sources such as websites, web pages and PDF files.
Learn how to crawl dynamic data sources: APIs, databases and so on.
Learn how to crawl dynamically loaded data and specific page sections with the Puppeteer crawler.
Understand how Alan AI crawls documents and what the depth
parameter means in different types of corpuses.
Learn how to prioritize one data corpus above the others.
Learn how to filter data corpuses based on criteria like user roles, product versions or individual preferences.
Learn how to include and exclude specific documents from data corpuses.
Find out how to crawl websites and web pages that require basic authentication.
Discover how to manage data crawling tasks in Alan AI Studio.
Learn how to review and examine what data sources and content your Q&A AI agent utilizes to converse with users.