Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/arainey2022/myaskai-docs/llms.txt

Use this file to discover all available pages before exploring further.

Self-learning is one of the most impactful features in My AskAI. Rather than requiring you to manually update your knowledge base every time a gap is discovered, the agent analyzes conversations where a human agent stepped in and automatically generates new knowledge articles to fill those gaps. It happens in the background, on a weekly basis, without any manual effort on your part. Accounts with self-learning enabled have seen a 40–60% drop in previously unanswerable questions and a 5% increase in AI resolution rate — with the biggest gains occurring immediately after enabling the feature.

​
How Self-Learning Works

Self-learning identifies the difference between what your human agents know and what the AI currently knows — then bridges that gap.
1

Conversation Is Escalated

A customer’s question couldn’t be answered by the AI, so it was handed over to a human agent. The human agent replies and resolves the ticket.
2

AI Analyzes the Gap

My AskAI compares the human agent’s reply against what the AI agent knew at the time. It identifies the missing knowledge that would have allowed the AI to answer.
3

Knowledge Article Generated

An AI-generated article is automatically drafted containing the generalizable knowledge from that human reply. Only knowledge that applies broadly is captured — not user-specific information like account details.
4

Article Published to Knowledge Base

The article is automatically published and made available to the AI agent. From that point on, the agent can answer similar questions without escalation.
5

Weekly Updates

Every week, existing self-learning articles are automatically updated using new human agent replies from the previous seven days. You receive an email notification summarizing which articles changed.

​
Reviewing Self-Learning Articles

You can see everything the agent has learned — and review or edit any generated article — from your dashboard. Navigate to Dashboard → Improve → Self-Learning to see the full list of AI-generated articles. Articles that were updated in the latest weekly run will have a green dot next to their title. Clicking into an article shows additions highlighted in green and deletions in red, so you can see exactly what changed.
You can edit any self-learning article at any time. Articles use Markdown formatting, and you can preview how the content will appear before saving. Use the tone of voice settings in Self-Learning → Settings to ensure generated articles match your brand voice.

​
Self-Learning Modes

Self-learning works in both of My AskAI’s reply modes:

Direct Reply Mode

In direct reply mode, self-learning analyzes any conversation where the ticket was escalated to a human. It compares the human’s replies against the AI’s responses to identify the knowledge gap.

Notes Mode

In notes mode, self-learning reviews the entire conversation — both human and AI replies — to find gaps. This means your agent is learning and improving even while you’re still testing it before going live.
Self-learning only captures generalizable knowledge — information that would be useful for answering future questions from other customers. It will never learn from user-specific information such as account details, personal data, or one-off edge cases.

​
Managing Self-Learning

  • Turn off auto-publishing: Go to Improve → Self-Learning → Settings and disable publishing if you want to review articles before they go live.
  • Unpublish individual articles: Toggle the published/unpublished switch next to any individual article.
  • Delete an article: Open the article and click the trash can icon.
  • Change tone of voice: In Self-Learning → Settings, customize the tone of voice prompt used by the AI when generating new articles — use your own brand voice or generate a prompt automatically from your homepage.

​
The Improvement Feedback Loop

Self-learning is part of a broader feedback loop that makes your AI agent progressively better over time:
Self-learning works alongside the tools in the Improve section. Use Knowledge Gaps to see what couldn’t be answered, self-learning to automatically fill those gaps from human replies, and Guidance or Custom Answers to add further control where needed.
Every escalated conversation is an opportunity. Self-learning ensures that each one makes your agent more capable — automatically, at scale, without your team having to manually write documentation for every edge case your customers raise.

Build docs developers (and LLMs) love

Get started for freeTalk to us