COVID-19 : What we’re doing

We are doing our best to play our part by adapting responsibly and responding in any way we can. Over the last few weeks, we’ve launched a quick deploy Chatbot Capability for COVID-19 Information that is being taken up by a variety of organisations. Find out more here

The chatbot sector is currently in explosive growth. In 2019, 60% of business have now either created a proof of concept or have a chatbot fully running in their environment. For those looking to join the chatbot bandwagon, or take their chatbot to the next level, your choice of chatbot platform is a crucial decision to make. 

Today, we’re spoilt for choice of chatbot platforms. Which overall, can make your decision harder. (Too much choice paralyses people after all). 

So this article is here to make your life easier and save you a ton of time having to do the comparisons yourself: 

So which is the best chatbot platform? Overall, we found the best platforms are:

  • Proof of concepts & MVPs – Google Dialogflow 
  • Open source: RASA
  • Enterprise: Enterprise Bot Manager
  • Messenger Marketing: Chatfuel
  • Small business sales automation: Drift

Obviously, this is a very generalised summary and these may not be the best platforms for your personal requirements. 

What if you’re an enterprise that’s looking to do messenger marketing, and automate internal HR meanwhile wanting to have your platform on-premise? 

To answer such potential circumstances, this article will go in depth on each of the chatbot platform’s strengths and weaknesses.

It will also be worth reading these articles linked below, as ultimately, your chatbot platform choice will be affected by other external factors:  

What does it take to build a chatbot?
How much do chatbots cost to build

What is the difference between a platform and a framework?

A framework is a set of functions and classes which developers use for faster development. Typically bots will be built from scratch using programming languages.

Some examples of frameworks we cover here are Microsoft Bot Framework, Dialogflow and Botpress.

Platforms are online ecosystems where chatbots can be built by non-technical users and have a pre-built user-friendly interface. The depth and features of bot platforms vary widely. 

Why is it important to take your time to compare? 

Chatbots, natural language processing and AI are becoming a key component in any business. The likes of voice and chat assistants have become mainstream. Your business now has unprecedented access to communicating to your customer via hundreds of channels. 

As conversational AI continues to grow and become more commonplace, it’s important for businesses and developers alike to gear up to match the increasing customer expectations. 

Vendors range from the monolithic companies and cloud providers to specialised solutions and niche vendors that offer unique and powerful features. 

As we move into the slope of enlightenment and our chatbots begins to scale and grow in the number of uses cases and intents they can handle, it’s vitally important businesses double down and place their bets on their NLP platform of choice. This is because it will be increasingly difficult to change platforms as time goes on. 

As you go through the decision making process, there are multiple factors when considering your chatbot platform:

What are your data privacy & security requirements?

For every business, one of the most significant considerations is around sensitivity of your data. If your chatbot needs to process confidential or sensitive data, then you’ll need to question whether you want to upload this to a cloud based vendor. 

What is the level of confidentiality of data do you need to communicate through the chatbot?

What is the size of payload and frequency of API calls?

Another consideration if your payload (how big are the messages you’re sending) and the frequency of the API calls. 

With chatbots, there are a number of things that affect the API calls: 

  • Number of messages
  • Frequency of conversations / response rate
  • User message size
  • Chatbot response size
  • Images/gifs and other mediums.

With most of the platforms freemium models, it’s unlikely you’ll hit the upper limits of API calls you’re allowed to make. After a month or two of trialing, you should start getting data to make early estimates of what it will cost to scale.

We cover more points in the articles linked above. To summarise before we move on to the comparison: there is no one size fits all approach. If possible, it’s recommended you get professional experience when dissecting and using this document.

Comparison criteria:


  1. We try to be as unbiased as possible.
  2. The numbered gradings are entirely subjective to our opinions and are only there to provide a guide to performance instead of just giving you wall of text. 

We’ll break usability down into 5 sub categories to mark the chatbot platforms by: 


Business user-friendly:

Firstly, we look at the general user interface (UI) and user experience (UX) 

Asking question such as:

  • How beautiful is it?
  • Is it intuitive to use?
  • How easy is it to get started? 

Tying in closely to UI & UX and looking specifically at the business user, we’ll be looking to answer some key questions: 

  • Can a business person set up and maintain a conversational environment? 
  • How big is the learning curve? 
  • At what point do you need to get a developer involved to support the business user?
  • How good is it working with teams of different users and collaborating between them?

Developer Friendly

Here we take the same approach as analyse the usability of the platform, but from the developer perspective… 

Features highlights 

Here we’ll cover some of the features that we think make the chatbot platform stand out. If you’re a person with an eye for detail or want a comprehensive comparison list: our spreadsheet has full details of all the features that does just that. 

Build speed 

How quick can you build, make changes and deploy to live environments? We judge build speed from a combination of general UX, how much business vs developer time you need, how collaborative the platforms are; along with the pre-built entities and intents. 

2. Breadth of service

For many organisations, it makes sense to keep all your different functions under one provider. Many of our larger clients have huge contracts with AWS or Google cloud and makes everything easier to follow that theme.

To mention a few of the things we look for when comparing the breadth of service:

  • Does the platform provide an array of capabilities to enable computing solutions? 
  • Does it include:
    • Speech-to-text
    • Natural language processing
    • Text Classification
    • Natural language generation
    • Sentiment analysis 

3. NLP capability 

We created an article specifically around natural language processing where we analyse the major NLP providers in more detail here.

In this article, we’re more concerned about the entire framework, where NLP only makes up one part of the bigger picture. So to keep this article from turning into an entire book, we’ll summarise this into a score out of 10 and boil down these factors down into “F1-score”.

F1-score is the overall score from a combination of precision and recall of a test.

To quote from Wikipedia:

“In simple terms, high precision means that an algorithm returned substantially more relevant results than irrelevant ones, while high recall means that an algorithm returned most of the relevant results.”

F-1 score is the mean of precision and recall. 


Depending on your requirements, you may need your chatbot platform to support multiple languages. We find out what language each platform provides.

4. Integrations

Here we cover another critically important aspect to chatbot development: what the chatbot platform can integrate into. For example:

Front end integrations: 

These are usually channels where your customers can communicate with your business. 

Back end integrations:

Back en integrations are things like your customer relationship management systems and databases like:

5. Pricing

As always, pricing plays a big factor in your overall decision. Typically, pricing for platforms are by API calls. We’ll take a look at every factor of the chatbot platforms pricing model: from their freemium through to their premium and enterprise offerings. 


Here we’re looking to see how useful is their freemium model is and look for some key points: 

  • What’s the freemium features like? How much can you achieve with the freemium features?
  • How many hoops do you have to jump through to get said demo instance? Do you have to sign up for a demo and talk to a salesperson first or can you just sign up and go? 

6. Analytics

The correct success metrics, and therefore analytics, need to be in place before any chatbot project begins. Some things we’ll be looking out for: 

  • Does the platform provide analytics to monitor the health of applications? 
  • Does it provide feedback and remediation recommendations when users are experiencing issues?

7. Future product roadmap

What does the companies near-term (roughly one year) roadmap look like? What do they have in terms of in product enhancements, innovation strategy and partner ecosystem in the pipeline? 

Does the company have the resources and capabilities to deliver on its stated road map? 


Google Dialogflow

Breadth of service 9/10

Dialogflow is one of the most popular NLP tools and is the platform that powers Google Assistant that is used in over 400 million devices. Dialogflow was previously known as API.ai and was acquired in September 2016.

Google is one of the strongest contenders if you’re looking for AI capabilities. With their investments in Tensorflow and dedicated AI data centers such as their Tensorflow processing unit (TPU) for faster and more accessible machine learning training with AutoML

For example, Chatbase is a chatbot analytics platform born out of Googles incubator program: Area 120 – Chatbase gives you greater insight and analytics than Dialogflow alone.

Usability 7/10

Business user-friendly 6/10

Dialogflow has become one of the go-to platforms for natural language processing and building chatbots. It appeals to business users and developers alike. 

Following Google’s simple and minimalist design. One could argue its a bit bland to look at, however, It’s fit for purpose and gets the job done. Since Google keeps its enormous suite of platforms all separate, it enables an easy to use and intuitive user experience for business users. 

Its free and super easy to get started , especially if you have a google account, you simply sign in with that and you can get start in seconds.

The learning curve is fairly steep learning how to use Dialogflow. While the documents and training content availible is extensive, the tool does nothing to provide tips or tricks within the tool. Upfront training around conversation design, intents and entities is an absolute minimum.

Developers need to get involved quite quickly. The chatbots you can build without intervention from a developer are very basic.  Since any functionality beyond answering basic FAQs will need API integrations.
Lastly, there are no collaboration tools or features within Dialogflow. And one huge downside is the amount of duplication of work involved when it comes to converting your conversation design flows into Dialogflow. A seperate design tool is necessary to keep track of your conversation flows effectively. 

Developer friendly 9/10

Dialogflow has these current software development kits (SDKs):

  • Node.js
  •  Ruby
  • Android
  • iOS
  • C#
  • Python
Features highlights 8/10
  • Comes with extensive pre-defined ‘domains’ (knowledge packages), e.g. “small talk” or “flights” for understanding intents
  • Plentiful built-in integrations with key messaging platforms. For example, Slack, Facebook, Alexa
  • User-friendly tooling and intuitive interface for ongoing training and management
  • High performance – Dialogflow processes millions of user requests daily 
  • intuitive interface for ongoing training and management
Building speed 8/10

Thanks to Dialogflow’s pre built small talk and agent templates, you can get started very quickly. Dialogflow also has one of the greater collections of tutorials and resources that you can find online for guidance online. 

Natural language understanding 6/10

F Score
Languages – 10/10

Dialogflow has one of most extensive language support of all the NLU providers. To see their full language list, check here 

So much so, it’s a bit much to lay on the table the comparison here, as what languages are supported depend on a couple of factors, such as whether you want STT (speech-to-text, audio input, speech recognition) or sentiment analysis.

A full comparison chart with filters can be found here.

Integrations –  7/10

Dialogflow has open REST API capability if you need to write custom code.

In terms of built-in integrations, Dialogflow has 13.  

Pricing 6/10

Dialogflow has an average pricing model. No more expensive or cheaper than it’s competitors. 

Freemium 9/10

You can get a lot done with the free version of Dialogflow. It’s one of the better platforms for building proof of concepts (PoCs) and minimum viable products (MVPs).

The only constraining factor is that they charge per API call, which is plenty enough for small businesses and PoC. 

The enterprise edition offers responsive support and the ability to quickly scale as your business or application demands increase.

Limitations of the free model: 

  • Support is very limited 
  • No sentiment analysis 
  • Inability to scale quickly. 
  • A cap to how many calls you can make
  • A maximum document size of 10MB you can upload to knowledge connectors

Dialogflow pricing page


Dialogflow has all the basic analytics that you need. Filtering by dates, seeing successful intents and breaking down those intents into successfully understood

It also has our favourite to see analytics tool, session flows. Session flows are the visual summary of the conversational paths your users take when interacting with your agent.   

Future product roadmap

For Dialogflow, we can expect a continued expansion of in-built integrations with other platforms. Additionally, it’s inevitable Dialogflow will further integrate into Google’s broader cloud offerings.

Google is a strong contender for anything AI due to its extensive AI offerings and capabilities. 

For example, Google continues to make major investments in TensorFlow (a machine learning platform) and its data center capabilities like the TensorFlow Process Unit (TPU) for faster training.

Google is also expanding its device footprint with the likes of Google Assistant on Android and Google home.


Overall score 9/10

Dialogflow is your trusty chatbot steed. Its simple and easy to use interface gets the job done. The platform has extensive documentation, enabling you to get unstuck quickly. The number of prebuilt agents makes it really easy to start and allows for a relatively quick build time.
The future roadmap is seriously exciting (with the likes of AutoML and Knowledge store)

In a sentence: Dialogflow is superb for PoCs and level 2 chatbots and more than capable for all business sizes and projects.



Dialogflow is one of the most popular NLP tools and is the platform that powers Google Assistant that is used in over 400 million devices. Dialogflow was previously known as API.ai and was acquired in September 2016.

Breadth of service 3/10

Rasa has a framework that comes in three parts: 

  1. Rasa NLU: Rasa NLU (Natural language understanding) is seen as a popular alternative to the well known NLP services such as Dialogflow or LUIS. The key advantage of Rasa is that you have access to an entire Python pipeline and can integrate it with your back end systems with whatever complexity of custom logic you desire. 
  2. Rasa Core: Basically, core takes the intents and entities generated by the NLU and is the decision maker for what action to take. 
  3. Rasa X – sits alongside where you can annotate, get feedback, version control and manage your NLP models. 

So at the moment, Rasa is just about around creating conversations with NLU and dwarf compared to the likes of tech giants. 

Feature highlights
  • Provides high-quality speech recognition and language understanding capabilities, enabling creation of natural language ‘chatbots’ to new and existing applications. 
  • Integration with major platforms and APIs as well as having the flexibility to link to your own APIs and services. 
  • Open source and supported by a huge community of developers…’By developers, for developers’ 

Usability 4/10

Business user-friendly 0/10

The very fact you need to install Rasa through your terminal doesn’t bode well for your business user.

Rasa operates entirely through code and is currently not something accessible to anyone except developers. So let’s quickly move onto… 

Developers 10/10

Made by developers for developers. Rasa is the hardcore NLP platform for chatbot developers. 

Integrates with your APIs, pulling and pushing data to your enterprise backend systems. If you don’t want a standard connector to messaging channels like Facebook, Slack or Telegram, you can easily add your own.

Build speed 2/10

Rasa doesn’t provide any pre-built agents or intents compared to the likes of Dialogflow or Watson. However, since Rasa is focused around customisation and high fidelity, this is the sacrifice you must make.

NLP Capability

Intent accuracy



The Rasa NLU can be used to support any language but this will be somewhat dependent on the specifics of the application using it.  

Integrations 4/10

Rasa doesn’t have any built-in integrations, which isn’t much of a problem as it’s designed for maximum customisation and flexibility. Every needs to be hand made! 

Pricing 10/10

Being open-source, Rasa is free to use for everyone. For the enterprise edition, you have to contact sale to get a custom quote. 

WIth the free/community edition you get: 

  •  View and annotate conversations
  •  Get feedback from testers
  •  Version and manage models
  •  Deploy easily on-prem or to your favourite cloud
  •  Community Support
Limitations of the free model

It’s only in the enterprise edition where you get:

  • Analytics
  •  Role-based access control
  •  Multiple deployment environments
  •  Single Sign-On
  •  Service Level Agreements (SLAs)
  •  Deployment and Installation Support
  •  Expert Support available

You can find the full breakdown of different features here


Rasa X is sort of an analytics tool where you can review and get feedback your chatbot conversations. But nothing in terms of the typical dashboard with data you’d expect. You will need to supplement Rasa with platforms like EBM or Botanalytics unless you pay for the enterprise edition, but we’re not sure what analytics you get and what that looks like.  

Future product roadmap

Rasa haven’t given much in terms of specific future features. The NLP and chatbot industry are all racing towards solving the same goals. Currently, the next hurdle is making contextual assistants easier to build and more effective at handling contextual conversations, so it’s safe to say Rasa, and every NLP platform will be looking to release their own version to solve this problem.

Summary 5/10

RASA is positioning itself as the open-source, on-premise and highly extensible alternative to incumbent platforms like IBM Watson.

If you have high data security and privacy requirements or prioritise having your NLP on-premise, Rasa is the leading option for you. 



Microsoft LUIS (Language Understanding Intelligence Service) is a machine learning-based service for building natural language understanding into apps, bots and IoT devices. 

Developers will love Microsoft’s conversational solutions. While Microsoft splits the entirety of its solution across three product areas — the Microsoft Bot Framework, LUIS for natural language processing, and intent management and Azure Cognitive Services for extended AI support — it brings them together in one strong development environment.

  • Microsoft is strong in the enterprise but less so in consumer and devices. While Microsoft does play in the consumer device market, it will lag more dominant players like Amazon and Google and hence be less relevant in this market.
  • Microsoft will be a dominant enterprise cloud player. With its dominant position in knowledge worker technologies such as those in Office 365, insights into worker behaviour, and strong cloud development capabilities in Azure, Microsoft presents a very strong vision, especially for enterprise knowledge workers.
  • LUIS
  • LUIS the natural language understanding part of the Microsoft X suite. 
  • The platform uses intents and entities. 

 All LUIS active learning applications are centred around a domain-specific topic or content related. You can use pre-existing, world-class, pre-built models from Bing and Cortana. Deploy models to an HTTP endpoint with one click. LUIS returns easy-to-use JSON. LUIS stores all incoming expressions in the Logs section and provides semi-automatic learning features with Suggestion, when the system tries to predict the correct intents that are already present in the Model.


Business user-friendly 4/10

The bot services feels very…microsoft. Not the most intuitive to use or the nicest to look at, but functional.

The user experience is terrible to use compared to other platforms due to the endless navigation of senseless layers. 

All of the documentation is on Github and is hard to navigate unless your familiar with the platform. 

  • How beautiful is it?
  • Is it intuitive to use?
  • How easy is it to get started? 
  • Can a business person set up and maintain a conversational environment? 
  • How big is the learning curve? 
  • At what point do you need to get a developer involved to support the business user?
  • How good is it working with teams of different users and collaborating between them?
Developers 9/10 

Of the three services tested, LUIS has the weakest documentation and samples making the use of its API difficult. 

Business users

User-friendly interface for inputting conversations and re-training. 

The interface is in English only.

Feature highlights 

  • Allows you to use pre-existing, world-class, pre-built models from Bing and Cortana whenever they suit your purposes and to build specialised models when needed
  • Draws on Microsoft technology for interactive machine learning and language understanding
  • This tool is able to configure Features. It helps to improve model and Intents detection. All you need is to add a “Phrase List” or/and “Patterned Feature.” 
  • Phrase list allows to define the list of similar objects (for example, names of cities or cuisine types), while the “Patterned Feature” enables to define a regular expression which helps your bot to recognize domain specific patterns (for example, flight numbers).
  • Activate your models on any device
  • Easy integrations with major platforms through the Microsoft Bot framework

Building speed 6/10

The bot framework has something called “skills” 

Breadth of service

Microsoft LUIS is part of the Microsoft Bot framework, which in turn is part of Microsoft’s azure bot services – a series of cloud-based machine learning APIs and services.

LUIS.ai is Microsoft Language Understanding Intelligent Service that was introduced by Microsoft in 2016. Nowadays, Microsoft provides several useful tools for bot makers. Besides LUIS NLP engine, tech giant offers Microsoft Bot Framework and Skype Developer Platform.

Microsoft Bot framework helps to build, test, and deploy bots for many well-known platforms such as Facebook, Skype, Slack, Cortana, Kik, Telegram, and SMS. Skype Developer Program, in turn, gives the opportunity to build apps for Skype.

LUIS.ai can be applied both with or without Microsoft Bot Framework. However, the combination seems to be a more reasonable option.

NLP Capability

Intent accuracy

The LUIS UI is always in English, but several languages are supported for understanding utterances / intents including:

  • English
  • German
  • French
  • Italian
  • Spanish 
  • Portuguese
  • Japanese
  • Chinese

Integrations 9/10

One of the strongest aspects of using the microsoft azure suite is the fact it has, practically, every service an enterprise can want is under one place. 

In terms of front end integreations, the bot framework connects with them all:
Facebook, Kik, Slack, Telegram, Twilio, Microsoft Teams, Skype, Web Chat, and email.


LUIS API — Free: 10,000 transactions free per month; LUIS API — Basic: Up to 10 transactions per second; $0.75 per 1,000 transactions.

Freemium 1/10

Unless you have a hotmail account, signing up for the Azure account is laborious and long. 

Compared to some other platforms where you create an account name and go – the azure suite requires you fill in every detail. Understandable and expected for an enterprise like Microsoft, but still a frustrating user experience.

Although the azure account gives you £150 “free” credit to use, you can’t actually enable the bot services without a subscription or talking to sales. 


Since by using microsoft bot framework you’ll be using the Azure cloud, the key benefit is you’ll be able to have all of your analytics in one place and you can take advantage of Power BI – one of the world’s most powerful analytics tools.

That being said, gathering particular chatbot data isn’t great, they don’t have flow charts like EBM or Dialogflow and you can’t break down intents into sub data sets such as X or Y either.

The training tool displays performance metrics on training but is very slow.


The overall experience of working with the bot framework is one of the worst of all the platforms we’ve compared.

This being said, LUIS is the highest performing in terms of F-score and NLP performance.  

However, the experience of working with LUIS from a developer perspective was the worst of all three platforms.

The free offering isn’t free at all, when when you do register, the free  budget is very restrictive, not even enough for a thoroughly tested proof of concept. 

However, LUIS is a good option for .NET developers, bot projects that require integration with enterprise software or where you want all of your in one place if you’re already part of the azure suite, then it may be more effective to use LUIS.

LUIS and the azure suite is also good fit for Cortana functionality. 



Business user-friendly 6 / 10

Lex has a user-friendly interface for inputting conversations and re-training. It’s structure is similar to other engines such as Dialogflow where you build conversations around the user intents and utterances. 

The utterances don’t have to be exact, but we found it’s worthwhile defining utterances with different slot variations. 


The sheer quantity of system slot type references is a bit much.

Offers APIs for all major programming languages & frameworks (including our preferred languages of Node and Python) with clear documentation and an active forum for support. Programmers have a variety of options for integrating Amazon Lex into their projects. Amazon provides SDKs for Android and iOS, which support text and speech input. Alternatively, text-only input can be processed using Java, JS, Python, CLI, .NET, Ruby, PHP, Go, or C++.

There is no import functionality which could be a disadvantage for large scale updates etc. 

Features highlights 
  • A huge set of prebuilt agents and skills: currently 30,000 and counting.
  • Amazon’s strong offering leverages AWS and Alexa’s prominence. Amazon also offers Lex for building customized voice and chat experiences on mobile devices and chat services. Developers like the breadth of capabilities as well as the ability to leverage AWS, particularly Lambda skills and investments.
  • Amazon Lex and Amazon Polly can be used together to create smart assistants for specific use cases, such as telephone support systems. Amazon Lex, itself, leverages AWS Lambda for intent fulfillment, Amazon Cognito for user authentication, and Polly for text-to-speech synthesis.
Building speed 9/10

Amazon lex has one of the largest pre-built libraries of all the chatbot platforms (30,000 and counting). They have blueprints, skills 

Breadth of service

  • To no surprise Amazon has a huge offering of services within the AI and machine learning spectrum with the likes of Lex, Lamba and Polly. 
  • Amazon will build upon AWS’ success in the enterprise. Alexa for Business will find success in the enterprise as a voice-controlled access mechanism. It will also have the edge in conversation-enabling new, AWS-based applications. It will leverage its integration capabilities to conversation-enable legacy applications.
  • Also released at the the same time is Amazon Rekognition, a deep-learning system capable of identifying characteristics of images, such as tagging objects, scenery, and faces. It can also identify the sentiments of faces (frown, smile, anger), and perform facial recognition.

NLP Capability


The AWS interface is in English only.

Overall the NLP engine supports 17 languages including:

English, French, Spanish, Germany, Japanese, and Portuguese, as well as support for regional accents and dialects (American, Australian, British, Indian, and Welsh English, Canadian French, Brazilian Portuguese, and American and Castilian Spanish).


As expected, Amazon Lex has tight integration to different Amazon services such as Lambda, Dynamo DB, SNS/SES, and others. In addition, the developers have provided connectors for popular SaaS applications like Salesforce, Microsoft Dynamics, Marketo, Zendesk, QuickBooks, and Hubspot.

By means of AWS Lambda, Amazon Lex can be integrated with the enterprise connectors Salesforce, Microsoft Dynamics, Marketo, Zendesk, QuickBooks, and HubSpot. For security purposes, all Amazon Lex transactions are communicated over HTTPS


$0.004 per voice request, and $.00075 per text request. Your usage is measured in “requests processed”, which are added up at the end of the month to generate your monthly charges.


The free model overall is one of the better enterprise platform experiences. AWS tries to personalise its suggestions and offerings depending on your details and information provided. 

Free offer — up to 10,000 text requests and 5,000 speech requests per month for free for the first year.

5 million characters of amazon Polly (speech to text)


Lex web interface gives you not only configuration features, but also a monitoring dashboard where you can review different metrics such as Text Request (Count), Speech Request (Count), Text Missed Utterances (Count), Speech Missed Utterances (Count), etc.


Amazon Lex is a well-supported platform that will continue to improve over time due to its deep learning capabilities. The ability to automatically scale is also a very powerful feature of Amazon Lex.

Although Lex is offered as a free trial initially, the monthly costs could become fairly high depending on the number of requests your bot is dealing with. 

Much like Microsoft, if you’re eager to have your business use Amazon Alexa, it’s worth considering building your bot separately on the Lex platform. Our personal opinion is that there are better options such as IBM watson or EBM as your basis infrastructure vs Lex.

IBM Watson

Breadth of service

Currently, IBM Watson is used by nearly 3000 companies and has been in production going all the way back in 2011 when Watson won Jeopardy on Live TV. After that successful publicity stunt, Watson expanded into an extensive amount of offerings. In fact, the largest amount of all the providers we cover.

In terms of chat and voicebots, Watsons functions can be broken down into:

Google is one of the strongest contenders if you’re looking for AI capabilities. With their investments in Tensorflow and dedicated AI data centers such as their Tensorflow processing unit (TPU) for faster and more accessible machine learning training with AutoML

For example, Chatbase is a chatbot analytics platform born out of Googles incubator program: Area 120 – Chatbase gives you greater insight and analytics than Dialogflow alone.


Business user-friendly

IBM’s isn’t the most intuitive bot builder, or the nicest to look at. However, with its 15- to 30-minute chatbot tutorials, you can grasp it with relative ease and build your first chatbot without any developer intervention. 

 first teaches users how to quickly create intents and entities, which can create meaningful examples for Watson to understand and train with.


SDKs in Python, Java and Javascript make integration easy. The API for managing intents and entities is simple and intuitive.

Features highlights 

Building speed 8/10

You can build a extensive FAQ chatbot very quickly thanks to a couple of Watson assistant features:

Firstly, it has a large catalogue of agent templates and a suggestion generator. Intents and entities are fairly quick to set up thanks to its business user-friendly interface. Intents can also be imported from transcripts or from Watsons extensive content catalogue. 

NLP Capability

Intent accuracy


Watson Assistant supports the below languages. Some features are limited to Beta modes so we recommend confirming the level to which Watson supports each language. More information on language support for Watson

  • English 
  • Arabic
  • Chinese
  • Chinese
  • Czech
  • Dutch
  • French
  • German
  • Italian
  • Japanese
  • Korean
  • Portuguese
  • Spanish


Pricing: Lite, Standard and Premium plans are available.

  • 10,000 API calls free per month*
  • Up to 5 workspaces (chatbots)
  • Up to 25 intents
  • Up to 25 entities
  • Shared public cloud
  • $0.0025 (USD) per API call*
  • Unlimited API queries/month
  • Up to 20 workspaces
  • Up to 2,000 intents
  • Up to 1,000 entities
  • Shared public cloud

Watson Premium plans offer a higher level of security and isolation to help customers with sensitive data requirements upon request.


In the lite version, testing out Watson is also easy and completely free, allowing users to make up their minds about the product before purchasing it.

Consumers who want to test out the capabilities of IBM Watson-developed chatbots can open a free account capped at 10,000 API calls.

This free “Lite” account also features five skills, pre-built content, a seven-day analytics dashboard, and 100 dialogue nodes.

Users can learn how to create a chatbot from scratch through the functionality provided by IBM Watson Assistant.

IBM’s 15- to 30-minute chatbot tutorial first teaches users how to quickly create intents and entities, which can create meaningful examples for Watson to understand and train with.

It then demonstrates how to capture context from users to reduce redundancy by using slots.

Finally, it shows how to structure and organize dialog flows through the use of handlers and digressions, tool that can manage users that go off-topic and lead them back to their original conversation, respectively.


The platform’s analytics dashboard and conversation logs help users visualize and understand trends in user message data.

This can lead to action on any messages that might have led consumers to confusion.

Watson Assistant can also help avoid any conflicts that may arise if multiple teams create sample sentences for the same virtual assistant.

The Watson AI automatically recommends fixes that can be done to resolve this issue.

Over time, the conversations stored by Watson can become a repository of data regarding customer preferences and modes of engagement.

The ability for Watson’s chatbots to learn can significantly increase the impact of a business.


This platform is powerful but doesn’t seem friendly enough. On one hand, there are many building blocks that you can use in your application in addition to the Dialog API available in the Watson Assistant interface. On the other hand, you’ll have to spend much time to integrate them into your project.

IBM Watson is a big player in the AI/ML space and Assistant is a good product. However, in general, Watson services can end up being an expensive option.

It can be hard to manage large, complex bots using the Assistant tooling.

Watson Assistant tool requires some effort to start working with it and take advantage of its integrations. It’s an enterprise-level solution, and it doesn’t sound like an option for an MVP chatbot project.

Enterprise Bot Manager © 2020 Filament Consultancy Group. Registered in England and Wales -  Company Number 10180537.  © 2020 Filament Consultancy Group Canada Limited. Registered in Ontario, Corporation Number 1995332.