The 21 best AI chatbots of 2024

AI chatbots aren’t a luxury anymore – they’re the standard for providing an exceptional customer experience. According to the Zendesk CX Trends Report 2024, 67 percent of business leaders understand that chatbots can help build stronger customer relationships. As we learn more about the benefits of chatbots for businesses and customers, choosing the right AI chatbot is more important than ever.

You may have heard about AI chatbots thanks to OpenAI’s launch of ChatGPT in 2022. While ChatGPT is certainly one of the most popular conversational, generative artificial intelligence (AI), it isn’t purpose-built for every use case. Our guide details what you need to know about AI chatbots and ChatGPT alternatives for business and personal use in 2024.

• Popular AI chatbots at a glance
• What is the best AI chatbot right now?
• The 21 best AI chatbots
• A comparison chart of top AI chatbots
• How to choose the right AI chatbot software
• Frequently asked questions
• See why our AI agents are the best AI chatbots for CX

• General use chatbots: ChatGPT, Perplexity, ZenoChat
• Customer experience and employee service chatbot: Zendesk
• User experience chatbot: Gemini
• Web search chatbot: Microsoft Copilot
• Creative writing chatbot: Claude
• Sales and marketing chatbots: Jasper Chat, Copy.ai, ChatSpot
• Gaming and fantasy chatbot: Character.AI
• Research and collaboration chatbot: Chatsonic
• IT and HR support chatbot: Workativ
• Education chatbots: Khanmigo, Socratic
• Healthcare chatbot: Woebot Health
• Personal chatbot: Pi
• Open-source chatbots: Llama 3, HuggingChat
• AI chatbot aggregator: Poe
• Social media chatbot: Grok

ChatGPT stands out as the best AI chatbot for generic use cases due to its versatile conversational abilities, natural language understanding, and adaptability across diverse domains. Zendesk, on the other hand, excels at customer experience (CX) chatbot use cases. It’s purpose-built for CX, having trained its customer service chatbots on billions of customer service interactions, so it’s ready out of the box.

1. ChatGPT

Use case: General use

Models: GPT-3.5, GPT-4, GPT-4o

When OpenAI released ChatGPT in late 2022, it became the fastest-growing consumer app ever. It arguably popularised artificial intelligence, making generative and conversational AI tools accessible to everyone.

ChatGPT offers writing, summarising, coding, research and parsing capabilities. ChatGPT can also integrate with add-ons and plug-ins to access accurate information or use third-party services. Some notable plugins include:

• OpenTable for restaurant reservations
• KAYAK for travel bookings
• Instacart for local grocery shopping
• Speak for learning other languages

Since its launch, ChatGPT has rolled out new iterations of the original intent model, such as GPT-3.5 (available for free plans). GPT-4, which includes additional performance capabilities, is accessible starting at $20 per user per month.

ChatGPT also offers DALL·E, an AI-powered image generator, in ChatGPT Plus and Enterprise. Users can enter a prompt or describe an image through voice chat. Based on the description, DALL·E will generate an output, such as a picture, logo or comic strip.

ChatGPT has a few noteworthy limitations. The GPT 3.5 data set doesn’t extend past the end of 2022, so some information may not be current. It might lack real-world knowledge and struggle with understanding context, leading to occasional irrelevant responses. Additionally, it can be susceptible to generating biased or inaccurate responses when prompted to do so.

Pros and cons

Recommended reading: Learn how ChatGPT-4 integrates with Zendesk.

2. Zendesk

Use case: Customer experience and employee service

Models: proprietary

Unlike many AI chatbot solutions, Zendesk AI agents, the next generation of AI-powered bots, are purpose-built for CX and perform like human agents. They come pre-trained on over 18 billion real customer service interactions specific to your industry, so they already understand the nuances of the customer experience and your customers’ needs. This allows the system to deliver value on day one, saving your team time and avoiding the costs of manual setup.

Zendesk AI agents are equipped to understand and resolve even the most complex customer requests from start to finish. Taking it a step further, Zendesk enables teams to:

• Use generative AI to swiftly construct a knowledge base, enabling the launch of an AI agent within minutes.
• Accurately identify areas where customers seek additional support and construct intuitive conversation flows that systematically guide them toward ideal resolutions.
• Seamlessly address customer enquiries from initiation to completion by integrating with backend systems through APIs, facilitating data exchange for informed AI agent responses.
• Leverage customer insights to optimise AI agent performance and extend capabilities to new applications, channels, languages, and markets so you can quickly scale to meet growing needs.

Beyond conversational bots, Zendesk AI also holds benefits for agents. Our AI-powered agent tools include intelligent routing and triage—sending requests to the right human agent based on language, intent (what a conversation is about), and sentiment (if it’s positive, neutral, or negative)—and sharing these insights with agents to provide helpful context and suggestions for how to solve the issue.

AI can surface similar tickets, turn a few bullet points into a full reply, change the tone, and summarise conversations to boost productivity. The Zendesk agent copilot can take the pressure off agents by proactively offering reply draughts or next steps during customer interactions.

What’s more, Zendesk AI agents automate up to 80 percent of support requests, leveraging company knowledge and fully customisable conversation flows to resolve even the most sophisticated use cases. With Zendesk, companies get the flexibility and control to deliver customer support their way—whether through fully autonomous AI agents, workflow automation, or human touch, starting at $19 per agent per month. See for yourself with a 14-day free trial.

Explore more Zendesk pricing plans.

3. Perplexity

Use case: general use

Models: GPT-4o, Claude 3, Llama 3, proprietary, and more

Perplexity is a general knowledge chatbot. It uses advanced natural language processing (NLP) and large language models (LLMs) to understand user queries and provide sources and citations to back up its responses.

Its search engine uses generative AI, including models from OpenAI and Meta’s Llama. The conversational interface allows users to ask questions and get natural responses. Academic or professional researchers can use the source citations included in responses to double-check accuracy and include them in their papers.

Perplexity also features Visual Perplexity, which allows the chatbot to respond to users using pictures and videos, and Llama Chat, an AI chatbot built on Meta’s Llama 3 model to support common programming languages and help developers with code.

Additionally, perplexity features Copilot, a GPT-4-powered AI assistant. Users with a free account may only input five Pro queries per day, along with unlimited quick searches, while paying users get 600 Pro queries for $20 per month. Paid-plan users can also use the “Quick Search” function more freely. It’s not powered by GPT-4 and has more rudimentary capabilities, like serving up basic answers and search results to a question.

Perplexity.ai has its fair share of limitations and may occasionally generate factually inaccurate results. So, you might also end up with sentences that sound good statistically but include wrong information. Perplexity.ai may have issues understanding nuances of human language, such as sarcasm, humour, and cultural context, which can work for academic use cases but isn’t as effective for casual conversations.

4. Gemini

Use case: User experience

Models: Gemini, Pro 1.0 and Ultra 1.0

Gemini (originally Bard) is a conversational, generative AI chatbot developed by Google. Gemini’s capabilities include engaging in multimodal dialogue (understanding and generating responses using different forms of communication such as text, speech and images) and comprehending and producing content in multiple languages. It can also understand cultural nuances, enabling communication across different societies.

Gemini can complete tasks like creating games, solving visual puzzles and generating images with accompanying text descriptions. Its AI model is capable of logical and spatial reasoning – meaning it can follow instructions and understand logical relationships and spatial configurations, like in problem-solving tasks or game-play scenarios.

Gemini comes in four model sizes: Nano, Flash, Pro, and Ultra. Gemini Nano is optimised for smartphone and mobile devices, specifically for the Google Pixel 8. It can perform tasks like generating chat responses and summarising text. Gemini Pro operates from Google’s data centres, delivering fast responses and handling complex queries. Gemini Ultra is Google’s largest model, designed for highly complex tasks, coding, logical reasoning and creative collaboration.

Gemini Advanced enables detailed conversations and understands more context than its previous versions. Gemini can serve as a personal tutor, generate step-by-step instructions and assist with advanced coding scenarios. It can also analyse trends and help content teams brainstorm and create new content.

Google offers a free version of Gemini with the Pro 1.0 model that helps with common enquiries and integrates with Google apps. For $19 per user per month, users can access the Ultra 1.0 model that targets more complex tasks and is available in Gmail, Docs, Slides, and Sheets (for Google One AI Premium subscribers).

5. Microsoft Copilot

Use case: Web search, chat, image generation

Models: GPT-4 and GPT-4 Turbo, Dall-E

Microsoft Copilot (formerly Bing Chat) is an AI-powered “companion” for web and mobile applications. With Copilot, users can ask questions and engage in follow-ups, get article summaries with personalised explanations and generate emails and images. Copilot can also perform routine tasks with AI assistance, like suggesting meal plans.

Although some say it’s less conversationally creative than its competitors, the average user may find it handy for general planning, composing responses, and generating images.

Chatbot users can also view AI-powered results using the Bing search engine or app but have to download Microsoft Edge to get the full Copilot conversational experience. Copilot has a visual search and an enterprise-level chatbot that offers security features and citations for the answers it provides.

Microsoft Copilot has a free version, but there’s one condition: users must access the chatbot through the Microsoft Edge web browser. If the user has other browsers like Chrome, they’ll be restricted to 5 replies per conversation. Additionally, users can only engage in 300 conversations per day, with each thread capped at 30 replies, limiting extensive back-and-forth interactions. For $20 per user per month, users can get priority access to GPT-4 Turbo, along with other features.

6. Claude

Use case: Creative writing

Models: Constitutional AI (CAI), Claude 3 and 3.5

Claude is a generative AI chatbot from Anthropic that uses Constitutional AI (CAI). The response bot software refers to itself as “self-contained”, meaning it responds to general queries without searching the internet. Claude uses the HHH system architecture, which stands for “helpful, honest and harmless”.

Users can modify Claude’s behaviour by prompting it with background knowledge to receive the desired responses. Along with the free plan, Claude has two paid plans—Pro and Team—starting at $20 per person per month that can help users with text analysis, summarisation, and creative content generation. The paid plans provide access to additional AI models.

Since its original release in March 2023, Claude has upgraded to Claude 3.5 Sonnet, which was implemented in June 2024. Claude 3.5 Sonnet improved over previous models, solving more problems and responding faster than the previous Claude 3 Opus. Claude can visually analyse images, like photos and handwritten notes, and transcribe them to text. It can also code in multiple languages, so users can create and modify websites.

Claude offers a range of capabilities but with some limitations, like daily usage limits. Despite its conversational abilities, Claude is not a substitute for human intelligence. If a user asks it to summarise a web link, it may make up information.

7. Jasper Chat

Use case: Sales and marketing

Models: GPT-4, Anthropic, Google, and proprietary models, and more

Jasper Chat is an AI content and copywriting chatbot for businesses. Jasper’s AI bot helps brands generate content for various mediums, such as social media, advertising, articles, emails, and websites. It also helps users with ideas, maintains tone consistency, adheres to editorial guidelines and company information and streamlines campaign collaboration.

Jasper can generate various content types, including video scripts and blog posts. It can also generate content and drafts based on bullet points or conversations. To use this tool, the user enters a prompt, refines their request and exports the final result. Businesses can train Jasper on their preferred datasets and retain ownership of its outputs.

Jasper’s AI bot ensures content adheres to a brand’s voice and style while providing access to background information about the company for factual accuracy. It offers suggestions for content improvement and automated project management, enhancing transparency and efficiency in content generation tasks.

Additionally, Jasper can provide AI-generated text directly within Google Docs. It also offers AI recommendations for analytics and insights to enhance user performance and optimise content based on usage data. Plans start at $39 per month, and users can test them with a seven-day free trial.

8. Workativ

Use case: IT and HR support

Models: GPT-4

Workativ is a conversational AI platform that provides an AI chatbot to automate workflows and IT support for employee issues and requests. The generative AI-powered chatbot Workativ Assistant helps employees handle issues independently without involving an IT support agent.

Workativ Assistant can understand the context of an enquiry and respond with relevant answers to facilitate self-service. It helps with unlocking accounts, reporting issues, password resets, access provisioning, account updates, email verification and employee processes like onboarding and offboarding. The platform leverages Knowledge AI, powered by LLMs and generative AI, to enhance the knowledge base and respond to user queries.

Workativ also provides a no-code chatbot builder that enables users to create customised bots. These bots can manage conversations, answer FAQs and integrate workflows. They can also notify users via chat about upcoming tasks, like reminders about expiring passwords, incomplete surveys or personal information updates.

Workativ ensures the secure handling of user information provided to the bot, allowing admins to resolve user queries without storing or displaying sensitive data. For example, when users want to reset their password, they can provide the new password to the chatbot, which updates the password without storing or displaying it.

The platform also offers dynamic notifications to proactively notify users about actions they need to take in the workplace, such as updating passwords or filling out surveys. Users can also set up notifications using app triggers, providing endless possibilities for engaging with employees.

Workativ provides analytics and logs for insights into chatbot interactions, allowing users to monitor performance easily from a single dashboard. This includes tracking chatbot usage, sessions, user queries and live status. Users can also review chat history and past conversations to identify areas for improvement and enhance chatbot effectiveness. Users can start using Workativ for free with limited features or purchase the Starter plan for $1,530 per month.

9. ChatSpot

Use case: sales and marketing

Models: GPT-4 and Dall-E

ChatSpot is HubSpot’s free AI-powered assistant that combines ChatGPT with HubSpot CRM data. Though the chatbot includes access to HubSpot, you don’t need to use the customer relationship management (CRM) software to use the AI support bot.

The public beta version of the bot lets marketing and sales teams draught emails, write social media captions, generate blog topics, and develop calls to action. ChatSpot can use prompts to surface research insights, keyword rankings and answers to common questions. ChatSpot can also suggest prompts and templates for users to input into the chat. It comes with prebuilt sales prospecting and marketing templates, too.

The tool allows users to summarise CRM data, generate reports and analytics, conduct keyword research, manage customer contacts, generate follow-up correspondence, write blog posts, and post directly to HubSpot websites.

Additionally, the AI chatbot can collect company data and competitor analysis. With access to ChatGPT, ChatSpot offers additional writing functionalities, which help users create communication and marketing materials.

ChatSpot integrates with Google Drive, enabling users to send prompts directly to Google Docs, Sheets or Slides to generate content. It also integrates DALL-E 2 to provide reports in various formats.

10. Chatsonic

Use case: Research and collaboration

Models: GPT-4, Claude 3, Stable Diffusion and Dall-E 3

Chatsonic is a generative AI chatbot designed for content creation. Chatsonic can generate content directly from the chat window to various platforms like blogs or social media channels. It uses two of OpenAI’s intent models, GPT-3.5 and GPT-4, to enhance conversational experiences.

It has voice-to-text and text-to-voice capabilities that allow users to interact with the AI through spoken prompts. Users can request digital art outputs or content of any length, whether that be captions, email replies or long-form articles. Chatsonic also offers Chrome extension plug-ins to make it easier for users to write and research by assessing and fact-checking information about events and topics in real time. That way, users are more likely to receive accurate results during the research process.

Users can share conversations with others, enabling collaboration and feedback. Image generation using models like Stable Diffusion or OpenAI’s DALL·E expands Chatsonic’s capabilities beyond text, allowing for visual content creation. A prompt library offers premade templates for generating content, including SEO keywords. Users can save their prompts for future use or share them with team members.

ChatSonic also integrates with platforms like X and Slack to provide access to Chatsonic across different channels. Users can access limited features through a free plan or purchase Chatsonic for $12 per user/month.

11. Copy.ai

Use case: sales and marketing

Models: GPT-3.5, GPT-4, Azure, and Anthropic, and more

The Copy.ai chatbot is a text generation tool geared towards copywriting. The applications of Copy.ai include long-form articles, social media copy, product descriptions, content briefs, ad copy and sales copy. It’s designed for enterprise use cases in marketing and sales.

For example, the chatbot can write product descriptions for online shops, copy for digital ads, blog content and more. Copy.ai has unlimited brand voices, prebuilt prompt templates and an “Infobase.” It can also generate content in over 95 languages.

To get started, users must enter details about their project, including the topic, context and tone. From there, they just need to sift through the bot's outputs and select their favourite option. Then, edit, add more details if needed, and publish the new content on the platform of their choice. It also offers prompt templates to speed up content creation and a Brand Voice feature that analyses the content and infuses the brand’s voice, tone and style.

The platform’s free plan is accessible for occasional content generation needs. Copy.ai isn’t limited to a single AI model; it can work with various models like GPT 3.5, GPT 4, Azure, Anthropic and more.

Additionally, Copy.ai leverages web scraping to pull and incorporate information from the web so users receive relevant and up-to-date content. Copy.ai offers multiple user seats and shareable project folders for team collaboration. The free plan lets individual users access 2,000-word chats, while the Starter plan unlocks unlimited chats for $36 per user/month.

12. Socratic

Use case: Education

Models: Google AI and search technologies

Socratic by Google is a free search-based chatbot and learning app for education and research. It provides AI support for secondary school and university students to help them better understand their assignments. Socratic uses Google AI and search technologies to connect students with educational resources, including websites for study guides, tutorial videos on YouTube and step-by-step guides. It also uses text and speech recognition, so students have different ways to communicate what they need help with.

Once prompted with a query, Socratic shares a top match from Google and a detailed explanation, often with visualisations. The app also provides links to reputable online resources and study guides written by experts to enhance learning experiences.

While the Socratic AI chatbot by Google helps students tackle homework questions or understand complex topics, it does have its limitations. For starters, it mainly relies on information available online. So, it might provide outdated or inaccurate answers, especially for more niche subjects. Also, Socratic may not be able to provide the in-depth analysis you need for more complex or abstract concepts.

Another thing to consider is language support, which might not cover all languages or dialects, making it less accessible for some users. Like with any AI, there's always a risk of getting misinformation, so it's wise to double-check important facts. It may be best to use the tool to complement your own critical thinking and research efforts.

13. Pi

Use case: Personal assistant

Models: Inflexion-2.5

Pi is Inflexion AI’s free, personal AI chatbot designed for conversations, emotional support, and companionship. It responds with kindness, delicacy, and humour. Users can ask the bot for advice or answers to a particular query, brainstorm ideas, keep a journal, read a story or just vent. The bot also declares that it has no interest in passing judgement or offering unsolicited advice, allowing users to discuss more sensitive topics.

Positioning itself as possessing greater emotional intelligence than ChatGPT, Pi aims to engage users in friendly conversations while offering varied perspectives on multiple topics. The chatbot offers users unlimited queries and utilises the Inflection-1 language model. The human-like bot sends conversational messages (and even emojis) to emphasise its points. The bot is free to talk to, but you must make an account if you want Pi to remember the details you share.

Pi features a minimalistic interface and a “Discover” tab that offers icebreakers and conversation starters. Though Pi is more for personal use rather than for business applications, it can assist with problem-solving discussions. The Discover section allows users to select conversation types, such as motivational talks or venting sessions. Although Pi may not be the obvious candidate for boosting productivity, its focus on personal wellbeing sets it apart.

While Pi lacks the flexibility of ChatGPT for tasks like internet searches or article writing, it offers a user experience characterised by minimalistic design, animations and a preference for concise interactions. Pi fosters short bursts of conversation, often initiating discussions with open questions, like encouraging users to share their day or discuss personal challenges.

14. HuggingChat

Use case: Open-source

Models: Llama 3, CodeLlama, Falcon, Mistral, and more

HuggingChat is an open-source AI chatbot by HuggingFace. It functions similarly to ChatGPT, allowing users to craft texts, summaries, and content, debug code, formulate Excel functions, and address general enquiries. Users can start using the bot for free or upgrade to the Pro Account plan for $9 per month to unlock additional features.

It can help users create personalised meal plans and recipes or generate to-do checklists. The bot also elaborates on questions, predicting the user’s next question. If users have an account, they can refer to the thread for details later—even if they’ve made another query.

While the bot creates general content using its own data, you can toggle the “Search web” button so its outputs align more closely with other online results, giving you more recent information. Because it’s open-source software, users can access and modify the source code to customise the platform to fit their specific needs and add additional properties.

HuggingChat does come with a few limitations. It may produce inaccurate replies or show bias in its responses, and it is sometimes prone to generating hallucinations—an uncommon phenomenon that occurs when an AI chatbot provides responses not grounded in reality.

It occasionally stops generating output mid-response or strays from the original topic, particularly with longer prompts. While it’s useful for brainstorming, you may want to choose a chatbot that specialises in critical task generation.

15. Character.AI

Use case: Gaming and fantasy

Models: GPT-3

Character.AI users can choose the bot they want to talk to and the chatbot will reply using the tone and language one would expect from that specific character. The basic features are available for free, but users also have the option to purchase the Character.AI+ plan for priority service along with other perks. Here are some categories users can choose from:

• Helpers: dating coaches, psychologists, mock interviewers, and fitness coaches
• Famous people: Cristiano Ronaldo, Elon Musk, Albert Einstein, and Stephen Hawking
• Game characters: Mario, Luigi, Bowser, Nathan Drake, and Shadow the Hedgehog
• Movie and TV characters: Regina George, Captain Jack Sparrow, Captain America, Eleven, Wednesday Addams, and Tony Stark

Character.AI users can have entertaining “conversations” with their favourite stars and characters, individually or in a group. For example, users can have a one-on-one chat with Socrates, or a group chat with all the members of The Avengers. Users can also create their own characters and personalities and make them available for chats with other Character.AI users. They can even design bots for specific uses, such as a generative AI host that leads a text-based adventure game.

Character.AI chatbots do face certain challenges, such as requiring many resources to support large-scale simulations and occasionally getting stuck in repetitive loops. Users should be mindful of these limitations to manage expectations during interactions.

16. Poe

Use case: Chatbot aggregation

Models: GPT-4, DALL-E-3, Gemini Pro, Llama 2, Claude Instant, Claude 2, StableDiffusionXL and more

Its AI chatbots can answer questions, generate content, translate languages and play games. Poe’s native AI chatbot, Assistant, uses OpenAI’s GPT-3.5 model, allowing it to answer questions and generate content similar to ChatGPT.

Poe enables users to create customised chatbots with or without coding, accommodating several language models tailored to various applications like writing, role-playing and programming. Beginners can build a simple chatbot with its drag-and-drop tools, while developers can use Poe’s application programming interface (API) to integrate Poe with their current systems.

While training chatbots may take some proficiency, the platform features a Discover section. This tool lets users explore language models, tools for creating and managing chatbots, collaboration opportunities with other users and cross-platform accessibility.

Despite potential message limitations for premium models, Poe remains an accessible choice for exploration. Additionally, users can develop their chatbots, tailor prompts, integrate knowledge bases and monetise their creations through creator accounts, distinguishing Poe from OpenAI’s ChatGPT.

17. Llama 3

Use case: open source

Models: Llama 3

Llama 3 is a free AI model developed by Meta. It uses LLMs to complete tasks like text generation and code programming. With an open licensing framework, users can access some of the code, allowing them to customise the model to fit business needs (until reaching a high revenue limit). Bear in mind that access to Llama 3's development details is restricted.

While deploying Llama 3 is tailored for developers, users can experiment with it on the Llama2.ai website to understand its responses. The output is straightforward and less refined than other chatbots, providing a basic exploration platform with minimal customisation controls.

Llama 3 offers two model sizes: 8 billion and 70 billion parameters. The larger model can understand more complex data and concepts but requires more computing power and can be more difficult to use. Offering a variety of model sizes makes Llama accessible to businesses of all sizes. One of Meta’s goals for Llama 3 is to democratise the use of generative AI, making it available for personal, research, and commercial applications.

18. Grok

Use case: Social media

Models: Grok-1

Grok is a generative AI chatbot developed by xAI, the artificial intelligence company owned by X (formerly Twitter). The bot is named after the science-fiction novel “Stranger in a Strange Land”, and its persona is inspired by the comedic novel “The Hitchhiker’s Guide to the Galaxy”.

Grok uses real-time knowledge through the X social media platform to answer questions and suggest related follow-up questions for users to ask. It’s designed to answer questions with wit and humour, differentiating itself from other AI chatbots.

Although Grok’s access to real-time X posts reinforces its credibility, it’s also susceptible to inaccurate or unverified information. It lacks the safety guardrails of competing chatbots, allowing users to get answers about “spicy” topics that most other AI chatbots typically reject.

Grok’s capabilities extend beyond conventional responses. It can draft emails, debug code and generate ideas, all in conversational, human-like language. It can also handle multiple queries simultaneously. Despite its current limitation to text inputs, Grok is poised for future enhancements, including visual and audio capabilities.

The AI chatbot is currently in the beta stages and is only available for X Premium+ users for $16 per month. The xAI team that created Grok provides feedback forms for users to share their experiences. Grok's goal is to serve as a research assistant and data processor, helping users innovate with new ideas.

19. ZenoChat

Use case: general use

Models: ChatGPT-4 and Sophos 2

TextCortex’s AI chatbot software, ZenoChat, is a content generation tool for writing assistance, such as rewriting sentences or editing internal documents. It’s also ideal for translations, SEO, marketing support, information gathering and entertainment. It uses a natural language generation (NLG) tool and tailors each chat to the user with customised style, knowledge and communication needs.

The AI chatbot was trained using over 3 billion sentences to reduce plagiarism and create unique outputs. It also supports more than 25 languages, so users can communicate with people from different cultures and backgrounds.

The web search feature allows ZenoChat to provide the latest information from the internet. Users can customise their search by adding sources like Google Scholar, X (formerly Twitter), Reddit or customised URLs. Users can also customise AI personas and link knowledge bases ZenoChat bots can use during conversations.

ZenoChat features a marketplace with numerous prompt templates that enable users to browse and choose the task they want to complete. These templates guide users, helping them ask precise questions to get the best results. In cases where prompts are too brief, ZenoChat offers a feature that expands them to ensure the topic is suitably covered.

Users can customise the base personality via the chat box dropdown menu, toggle web search functionality, integrate a knowledge base or switch to a different language setting. In the free version, users are limited to 100 queries upon registration and 20 queries daily. Paid plans start at $5.59 per month and can be tested with a free trial.

20. Khanmigo

Use case: Education

Models: GPT-4

Khanmigo is an AI chatbot created by Khan Academy, an educational organisation. The AI-powered bot was designed to enhance learning experiences and provide personalised tutoring sessions. Khanmigo can provide teachers and tutors with effective strategies for teaching and engaging with students. Its virtual assistant helps teachers plan lessons and better understand their students’ needs.

For students, Khanmigo acts as an AI-powered, personalised tutor and can be used to help with assignments or break down complex topics. By leveraging the Socratic method, Khanmigo can help students find the correct answer without doing the work. It guides students through complex topics with thought-provoking questions and hints rather than simply giving them the answers.

Users can access interactive quizzes and review prior topics as they progress through Khan Academy’s curriculum. Khanmigo offers 24/7 access and leverages the GPT-4 language model for engaging conversations. Access to Khanmigo is currently limited outside the United States to certain English-speaking countries and covers a limited range of subjects, including art, history, and math. Khanmigo users can access the chatbot for free or pay $44 per year for additional features like career coaching.

21. Woebot Health

Use case: Healthcare

Models: proprietary

Woebot, an AI chatbot developed by Stanford psychologists and artificial intelligence experts, aims to support global mental health by engaging users in therapeutic conversations, mood tracking and interactive games. Founded in 2017 by Dr Alison Darcy, a clinical research psychologist, Woebot offers multiple clinical and therapy approaches, including:

• Cognitive behavioural therapy (CBT)
• Interpersonal psychotherapy (IPT)
• Dialectical behaviour therapy (DBT)

Woebot Health offers an AI chatbot that uses natural language processing and rules-based logic to provide 24/7 access to mental health support. Woebot’s conversational AI helps build relationships with patients, deliver personalised therapy techniques and monitor user progress. The bot can also refer users to human providers when necessary. Its non-judgmental interactions cater to different age groups, providing personalised guidance while ensuring data protection.

In addition to providing on-demand support, Woebot Health offers evidence-based cognitive behavioural therapy content, personalised care plans and mobile access.

While Woebot is free to use, it is currently only available to users in the United States, limiting accessibility. Despite its unlimited query capability, some users may find it repetitive, and its effectiveness varies from person to person. Additionally, the platform lacks human interactions, which may be a drawback for some users.

However, its therapeutic effect, mood-tracking feature, and interactive conversations make it valuable for enhancing well-being and engaging in therapeutic conversations.

The best chatbot platform depends on several factors, including your use case. Let’s explore some considerations to understand what to look for when choosing an AI chatbot for your business:

• Align use cases: Ensure its abilities align with your needs to get the maximum benefits. For instance, if a business prioritises CX, choosing a specialised chatbot would be ideal rather than an AI chatbot for sales.
• Integration capabilities: ensure the AI chatbot seamlessly integrates with your existing systems and software. Look for compatibility with your messaging software, websites, CRM systems, and other tools relevant to your business. Evaluate the availability of APIs and integration options that facilitate connections with third-party applications, enabling a cohesive workflow.
• Ease and speed of setup: Look for a pre-trained AI chatbot that works right out of the box. Determine if it requires developer support that may entail manual training and configuration, as this will delay its deployment and add to its costs.
• Scalability and flexibility: Scalability is crucial as it allows the chatbot to handle increasing volumes of interactions without compromising performance, ensuring consistent and reliable service as the business expands. Flexibility is equally important, as it enables the chatbot to be customised and updated to meet changing requirements, such as integrating new features or adjusting to shifts in business processes.
• Security: prioritise security features to safeguard sensitive information. The AI chatbot should comply with customer data protection regulations and implement encryption protocols to secure user interactions. Verify the platform’s privacy policies to ensure it handles user data responsibly and transparently. Look for options to customise data retention and privacy settings.
• AI model sophistication/LLMs: Know the foundational data the AI bot is trained on since this will influence its capabilities and effectiveness. High-quality AI chatbots often utilise NLP and ML algorithms to understand inputs and generate contextually relevant responses. Check for features like context retention, customer sentiment analysis and generative AI.

It’s also important to consider factors like selecting a chatbot with quality support that is regularly updated to offer the best user experience.