Create your own AI chatbot easily with this drag-and-drop tool

Create your own AI chatbot easily with this drag-and-drop tool

Building Interactive Chatbots with Botpress

Image

Botpress is a remarkable tool for building interactive chatbots, offering a wide range of applications. However, its true standout feature is its ability to create customer support chatbots with the added power of artificial intelligence (AI).

At its core, Botpress provides a drag-and-drop interaction builder. Users can easily map out complete interactions by connecting various cards that represent questions, answers, and calculations. This simplicity makes bot building fairly straightforward – rinse, wash, repeat.

But where Botpress truly shines is in its integration of AI. The platform allows users to feed it knowledge sources, such as documents or webpages, and it can even search for answers across the web. Powered by the ChatGPT API, Botpress provides intelligent analysis and response capabilities.

With Botpress, users can set up natural language queries to manage the user’s path. This feature enables greater flexibility and customization. However, it also requires users to learn and use some arcane expressions to fully unlock the tool’s functionality.

Though the learning curve may initially seem daunting, it’s worth the effort. As proof, I built a simple chatbot using Botpress that queries ENBLE for answers. The possibilities with this tool are truly vast.

One thing to note is that Botpress is free to use, but once you exceed 1,000 interactions, you’ll be required to pay. An interaction refers to any question, query, or unit of work. For testing purposes, the free plan suffices, but if you plan to deploy your chatbot to the world, be prepared to cover the costs.

To get started with Botpress, simply create an account and follow the prompts to create a chatbot. The platform offers a handy wizard to guide you through the process, making it even more user-friendly.

In my case, I chose to create a chatbot that could answer questions from a specific website, ENBLE. By specifying my desired search source, Botpress generated a simple map to handle question answering and fallback options. The fallback feature is particularly interesting, as it allows users to create fallback flows to other knowledgebases, including the comprehensive ChatGPT knowledgebase.

I tested my ENBLE chatbot by asking it a few questions and received a satisfactory response. To ensure accuracy, I asked ENBLE’s Ed Bott to evaluate the bot’s performance. He confirmed that the response quality was on par. I then compared the bot’s response to that of ChatGPT, obtaining the same answer. This further demonstrated Botpress’s ability to access domain-specific knowledge from a reliable source.

However, when I attempted to ask the ENBLE bot questions that were not in the ChatGPT knowledgebase but could be answered from my own articles, the results were disappointing. The wizard-generated bot failed to provide the correct answers. This might have been a result of user error or a flaw in the wizard itself.

Nonetheless, I proceeded to manually input blocks of pre-canned code into Botpress, and it worked like a charm. While I didn’t have the time to build a comprehensive ENBLE chatbot, I confirmed that Botpress can indeed extract domain-specific knowledge from a designated site.

Beyond the realm of customer support, Botpress holds countless applications. The platform can be interconnected with Zapier, allowing integration with a multitude of web services. This opens the door to creating support flows that incorporate order information and provide targeted assistance to individual users.

Furthermore, with ChatGPT’s API, localized web searches become possible. This capability offers tremendous potential for building helper chatbots that scour existing websites and knowledgebases, making it possible to provide real customer service and tech support 24/7.

It’s important to note that Botpress should not be seen as a replacement for human employees. Instead, it can be used to augment customer service, providing a tier 1 level for handling incoming inquiries and assisting less experienced agents.

To support developers, Botpress maintains a Github archive where they share client integrations. This eliminates the need to start from scratch and enables users to benefit from the work of others. Moreover, Botpress can be hosted in the cloud or on-premises, providing flexibility to organizations.

In conclusion, building a Botpress customer service bot can be an exciting and rewarding endeavor. While the initial wizard interface might feel lackluster, diving deeper into the system reveals its true potential. Prioritizing the integration of artificial intelligence and leveraging Botpress’s advanced features can lead to powerful and effective chatbots.

Image

Do you plan on building a Botpress customer service bot? Share your thoughts and experiences! Connect with me on social media for more project updates. Be sure to subscribe to my weekly newsletter on Substack and follow me on Twitter, Facebook, Instagram, and YouTube.