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.

This page answers the most frequently asked general questions about My AskAI — from how the underlying technology works to practical questions about plans, limits, and features. If you’re looking for questions about AI accuracy, see the AI & Accuracy FAQ, or for setup and integrations see the Integrations & Setup FAQ.
My AskAI uses a technique called Retrieval-Augmented Generation (RAG) combined with large language models. Here’s what happens behind the scenes when a user asks a question:
  1. Content ingestion — When you add a website, file, or document, My AskAI extracts the text and breaks it into chunks. Each chunk is converted into a numerical vector (an “embedding”) that represents its semantic meaning. These embeddings are stored in a vector database.
  2. Semantic search — When a user asks a question, that question is also vectorised. My AskAI searches for the six most semantically similar content chunks in your knowledge base.
  3. LLM answer generation — The question and the retrieved chunks are passed to a large language model (currently GPT-4o class models from OpenAI). The model is instructed to answer the question using only those chunks — not general world knowledge.
  4. Response delivery — The generated answer, along with source references, is returned to the user.
This approach grounds the agent’s answers in your content and prevents it from making things up or drawing on information you haven’t provided.
Your AI agent responds to text-based customer support questions. By default it will respond to any incoming message in your connected channels (Zendesk, Intercom, HubSpot, etc.) that arrives in a new conversation or ticket.Current limitations:
  • The agent cannot see images or attachments sent within a conversation. It will respond with a message asking the user to describe their issue in text. Support for image reading in live conversations is on the roadmap.
  • The agent is optimised for support questions. It is not a general-purpose chatbot — asking it where it’s from or requesting that it write a poem will produce a politely unhelpful response by design.
You can use Guidance to further scope which types of questions the agent responds to, or Playbooks (on supported channels) to control routing logic.
By default, no. My AskAI is configured so the agent only answers questions when the answer is contained within the content you’ve uploaded. If it can’t find a relevant answer, it says “I don’t know” and offers the user the option to speak to a person.This is a deliberate design choice for two reasons:
  1. Answers grounded in your content are verifiable — you can trace any response back to a source.
  2. Without this restriction, users could effectively use your My AskAI as a free general-purpose AI assistant, which would consume your conversation limits and pollute your Insights data.
While My AskAI trains its system carefully for factual accuracy, no AI service is 100% accurate. Always review answers to high-stakes questions.
Yes — but not automatically through passive conversation monitoring. My AskAI does not learn directly from individual conversations. Instead, it improves through deliberate human-in-the-loop actions:
  • Self-learning — My AskAI’s Self-learning feature surfaces conversations where the agent struggled, so you can review and improve them.
  • Knowledge Gaps — The Knowledge Gaps view shows questions the agent couldn’t answer, helping you identify missing content to add.
  • Custom Answers — Add Custom Answers for questions you want answered verbatim.
  • Content updates — Keeping your help docs, website, and knowledge sources up to date is the single most effective way to improve answer quality.
Over time, acting on these signals produces a measurably more accurate agent.
Yes. You can fully customise your agent’s name, tone, writing style, and behavioural constraints using Guidance. Guidance lets you add standing instructions that apply to every answer — for example:
  • “Always respond in a friendly, casual tone and sign off with ‘Let me know if there’s anything else I can help with!’”
  • “Never speculate. If you are unsure, direct the user to contact the support team.”
  • “Refer to yourself as ‘Aria’ and never acknowledge that you are an AI.”
Guidance is the primary way to shape the agent’s “personality” and ensure it reflects your brand voice.
Yes. My AskAI supports multiple agents via the Multibrand feature. You can have one agent per brand, product line, or customer segment, each with its own knowledge base, persona, and channel integrations.If you need multiple knowledge bases within a single agent (for example, to support multiple products under one brand), get in touch with the team via the chat widget to discuss configuration options.
Yes. There are several ways to bring existing macros or canned responses into My AskAI:
  1. Import Zendesk Macros — If you use Zendesk, go to Knowledge → Content and click Import Macros to convert them into Custom Answers automatically.
  2. Add Custom Answers manually — Go to Improve → Custom Answers and add trigger phrases with the exact response text. The agent will use these verbatim when a matching question arrives.
  3. Upload a document — If you don’t need exact verbatim responses, create a document containing your canned responses and upload it as a knowledge source. The agent will reference it (with natural language variation) rather than using it word-for-word.
For a large volume of macros from platforms like Freshdesk or Gorgias, contact support for assisted bulk import.
My AskAI’s AI agents scale with your usage and can process as many concurrent tickets as your plan allows. Some customers process over 100,000 tickets per month without issue.Each plan comes with a monthly conversation limit. If you regularly exceed your limit, you will see notifications in the dashboard and can upgrade your plan from Account Settings → Billing. For enterprise volumes, contact the team to discuss custom limits.The agent handles every conversation in parallel — having hundreds of simultaneous users does not degrade response times for others.
My AskAI (the current product) is a purpose-built AI customer support agent designed for businesses. It focuses on answering customer support questions across live-chat channels, ticketing systems, and email — and generates actionable insights from those conversations.My AskAI Classic is the earlier product, which was a more general-purpose AI chatbot builder for creating Q&A bots from your own content. It’s still available and supported.Key differences:
FeatureMy AskAI (new)My AskAI Classic
FocusCustomer supportGeneral Q&A
IntegrationsZendesk, Intercom, HubSpot, Freshdesk, Gorgias, etc.Embeddable widget
Insights & AnalyticsYesLimited
Self-learningYesNo
If you have an existing Classic subscription and want to switch to the new product, email team@myaskai.com. The Classic and new product use separate subscriptions and dashboards.
Yes. My AskAI offers a 99.99% uptime SLA over a 90-day rolling period for customers on the Enterprise plan.You can monitor current operational status and any ongoing incidents at the public status page: my-askai.instatus.com.
You can follow My AskAI and stay up to date with new features and case studies at:For the latest product updates, visit the changelog.

Build docs developers (and LLMs) love