liveBook · Manning

Conversational ai

As explained in AI for Everyday IT

Conversational AI is artificial intelligence that supports ongoing dialogue by interpreting user messages and generating context-aware responses.

Conversational AI lets people ask questions and work through tasks in dialogue with an AI assistant. A language-model chatbot generates responses to prompts and can help with practical work, from explaining an error message to drafting documentation. section 1.1 section 1.4.1

For an IT team, the useful starting point is a specific problem: describe what you need, provide relevant information, and ask for an output you can check. Depending on the assistant, you may be able to paste text, attach a file, or work through an integrated service. section 8.2.2 section 7.1.1

Find common VPN fixes

Attach a CSV export of service desk tickets and ask: “Based on the information in the file, provide the most common resolutions to problems with VPN connection drops.” An assistant might list VPN settings changes, client updates, and server switches, with counts for each. Check those claims against the tickets before choosing a fix. section 8.2.2 section 7.5

Choose an assistant for the task, not just its name. Some integrate with email or other workplace tools; others offer a more flexible general-purpose chat. Features and results vary, so try the same task with different services when appropriate. section 7.1.1 section 2.1.5

Table 7.1 Examples of AI-enhanced email management
Task
Prompt
Summarize an email conversation.
Find all email conversations over the past week from Manning.com. Summarize the emails, focusing on the main points, decisions made, and any action items or deadlines.
Respond to emails using AI.
Craft a response to the following email that addresses all questions asked, reiterates any important points, and proposes next steps.
Draft emails with AI assistance.
Help me draft an email to [recipient’s name/role] regarding [topic], ensuring it’s clear, concise, and professional.
Decode technical jargon in emails.
Explain the technical terms found in the attached email in layman’s terms, ensuring clarity and accessibility for non-specialist readers.
Check emails for tone, clarity, and professionalism.
Review my draft email below for tone, clarity, and professionalism, and suggest any necessary revisions.
Find unread or follow-up emails.
Identify all unread emails or emails that require a follow-up in my inbox from the past week.
Draft an email for various target groups on the same topic.
Draft a single email about [topic] that is suitable for sending to different target groups, including [group 1], [group 2], and [group 3], ensuring it resonates with each audience.
Tailor messages to your audience.
Adapt the following message for [specific audience], considering their interests, knowledge level, and cultural context.
Deal with technical jargon.
Rewrite the following sentences to remove technical jargon and make the information accessible to a general audience.
Make quick announcements.
Compose a brief announcement about [topic/event] that is informative and engaging for the entire team.
Provide feedback and suggestions.
Formulate constructive feedback and practical suggestions for the attached project proposal, ensuring the tone is encouraging and supportive.
Maintain inbox zero.
Provide guidance on effective strategies and best practices for managing a high volume of emails, with the goal of achieving and maintaining an organized, prioritized inbox that supports the concept of 'inbox zero'.

Use these email prompts as starting points for summaries, replies, and plain-language explanations. section 7.1.1

Keep these limits in mind:

  • Check the answer. Chatbots can make mistakes, and email searches may miss relevant messages. section 7.5 section 7.1.2
  • Watch the token limit. A long input can leave too little room for a useful response or result in important details being missed. section 3.2.2
  • Protect data. Follow workplace policies, remove sensitive details from prompts, and get security approval before using AI for anonymization. Avoid AI anonymization for critical or regulated data. section 7.5 section 6.8
  • Review for bias, especially when the output affects people, and follow applicable rules for disclosing or citing AI use. section 1.6.2 section 1.6.4

How AI for Everyday IT covers it

How deep it goes
Explains chatbot use through task examples, tool comparisons, and token limits rather than model internals.
Examples
Uses VPN ticket analysis, email summaries, and clarification questions for a fuzzy laptop screen.
What you can do after
Apply it via ChatGPT or another assistant to investigate IT problems and draft workplace communications, while checking results.
How it connects
Connects chatbots with generative AI, language models, APIs, and token limits.

Also explained in

Compare how 26 books explain conversational ai →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of AI for Everyday IT. The section links point to the parts of the book it draws on.