kore.ai
Published in

kore.ai

“I Can Never Build A Bot, I Don’t Have Enough Coding Skills!”, “I don’t belong to the chatbot domain, developing bots is not my forte”. “Bots must be developed by software developers only. Even if I know the use case, I don’t think I will be able to develop it”.

These are some of the stereotypes that we have often come across. Having the knowledge of coding is always an added advantage but is not a mandatory skill today to develop basic bots. With the advancement of bot platform features, the UI of the platform has become very simple and intuitive to enable the business users to create the bots by themselves. One such chatbot platform is provided by Kore.ai. The AI/ML features of the bot platform make the life of the bot developer very simple. Using this platform, the bots can be trained using the simple user conversations.

In this blog, I shall share my stint and learnings of developing a chatbot using kore.ai platform. If you think you or your team members perform a range of activities that are repetitive and does not add much value, then this blog is for you. You can now part with these repetitive activities by building a bot that can perform these tasks on your behalf. Don’t know much about coding? No worries! The platform will take care of it. You can create your chatbot in less than 15 clicks.! Yes, you got it right- 15 clicks or even less.. No coding required.!

Planning –

Before designing the bot., you should have a very clear idea of the use case. Make a user journey or story for your bot so that its functionality is clear. You can focus on the “What” and ‘for Whom” part. The “How” Part shall be dealt with later.

User Story: As a Manager, I should be able to automate the task of the fetching the documents so that my time of searching and accessing them manually is saved and, I can get the documents 24X7 without the assistance of the concerned team.

The documents management team (or asset management team) in general takes up the request to share the latest collateral with the sales, pre-sales or marketing teams. The collaterals include documents such as sales presentations, competitor analysis, case studies, whitepapers, blogs, videos, info-graphics, mark-toons and many more. Although the documents are always available in the shared drive, the team is apprehensive of fetching it directly. “I am not sure which is the latest version”, “I want the editable version only” There is a huge repository, I get lost in them “. These were some of the probable reasons why those team approached the documents management team.

The asset management team will find it extremely disadvantageous to provide similar kind of assets to multiple teams at different points of time. This is a common story for most of the enterprises. Even if they have a robust system, teams end up in asking for the assets from the content /product team only.

Approach –

Why not create a chatbot that can resolve this issue. The bot which understands the requirements of the various stakeholders and provides them with relevant and accurate solution.

In this use case, since the assets are limited and repeatable, a FAQ bot would be more appropriate.

A FAQ bot can be developed using Knowledge tasks in the task execution framework of Kore.ai platform. Knowledge tasks can be used to answer the most commonly asked questions freeing your support staff to focus on more complex questions. (read more)

The extensive feature of Knowledge extraction is used for this bot. All we have to do is spend some time thinking about what kind of queries the chatbot can get for the assets and use them to fill a simple the .csv template which is then imported to the platform.

The sheet would have the links to the drives which contained the latest and updated, editable and non-editable assets. I had used a zero-coding approach and develop the bot with just a few clicks. Sounds interesting so far!! Now let’s get our hands dirty and build the bot.

Our bot is just a few clicks away! Let’s get started

Building the bot -

The first step would be to create an account in the bot builder page of Kore.ai platform. You can try it out for free for a limited number of sessions.

Login(click#1) into your account and create a new bot (Click#2). You can create the bot in more than 20 languages by selecting the language in this section. Also, you can choose a specific type of bot depending upon the requirement. To know more about the type of bots refer the documentation.

Bot Builder Home page

For this project, we would require a standard bot only. So just give a suitable name to the bot (Asset Finder Bot) and click Create (#3).

Bot Summary Page

Welcome to the bot summary page. Don’t worry for now on the features and sub-features in this page as of now. Though the platform has more than 200 features, the UI is very simple and intuitive, just to meet the requirement of the users.

The left pane shows you the various features of the platform in accordance with the bot development life cycle. Click here to know more about every stage of the Bot development life cycle.( https://developer.kore.ai/docs/bots/chatbot-overview/koreai-platform/#)

Since we plan to develop a FAQ bot. We go to the bot task (click #4) and go to Knowledge Collection (#5).

Bot Tasks

In knowledge collection, we would extract the Q&A from the file (in CSV format). All we need to do is just create the knowledge task by clicking (#6) the import button on the Knowledge Graph Tab.

Click on Proceed (#7) and Browse (#8) option to get (#9,10) the file from your local system. You also have an option to download the sample CSV file for reference.

Here is a snapshot of the CSV file which has been imported for the bot. ( Click here to know more on the format to the CSV file for the Q&A).

Q&A to be imported to the platform

We can then trigger the import by clicking on the Next button (#11). Once done (#12), we can go to the Knowledge graph Tab to see how the ontology looks.

Bot Ontology

And we are almost there. Note that the nodes are automatically generated. We can see the primary and alternate questions in each of the nodes. We now must train the bot with these questions by clicking (#13) on the “Train” button.

Now brace yourself to test the bot. The bot might take 2 to 5 seconds to train itself with these questions.

Yes, it’s done! Your bot is developed.

You don’t believe it. Just talk to the bot you have now developed using the “talk to the bot” feature in the platform. Click (#14) on talk to the bot icon and shoot your questions.

This is a basic FAQ bot which we have developed. The bot answers to our queries on the documents which we are looking for. Just chat with the bot anytime and get the queries resolved.

What’s Next –

The next steps would be to make the bot more intelligent. The bot now is trained for only specific questions. But what if the user asks anything different or differently? Will the bot be able to answer? Well, it might not in all the cases. Though the platform is capable to respond to some of the phatic expressions, they also have to be trained to respond to the ambiguous and complex queries. So, the next steps would be to make the bot human like by enhancing its NLP capabilities. The next blog will cover in details on how we can enhance the NLP capabilities of the bot. Also, in the subsequent blogs, the bot developed shall be published using a suitable channel so that it can be made available to the users.

Data:

https://github.com/ANKIT081016/AssetFinderBot.git

--

--

--

Kore.ai is a market leading conversational AI partner for Global 2000 companies. Kore.ai specializes in AI-rich conversational solutions designed specifically for enterprises that aim to make digital customer interactions faster and more human, and employee work simpler.

Naturally.

An image of what a robot might look like if it snuck into the jellybean and licorice division. From the mind of a 1950’s human.

Reconstruct face with absolute dimensions from images with MICA

An Overview of Early Deep Learning Recommendation Systems

The spectrum of Wide & Deep models.

AI in Telecommunication Market Size, Share, Investment Opportunities, Top Regions, Growth &amp…

Some Ethics of Artificial Intelligence

Passing the Turning Point of AI Transformation

FDA Approves First AI Tool for Detecting Colorectal Polyps

Coming Soon: Modular Autonomous Vehicle Systems (MAVS) and AI Self-Driving Cars

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Ankit Shah

Ankit Shah

Product enthusiast and an avid reader https://www.linkedin.com/in/contactankitshah/

More from Medium

How AI is changing video editing

Counter Terrorism Use Cases for Cryptocurrency: Tracking Ransomware Payments Through Crypto

Rarible Clone Script — Create NFT Marketplace like Rarible

Getting past GIF stickers for IM and the case for personal AI — The Daily PPILL #54