The first step to resolving a user’s chat is to figure out the domain the user is chatting about.
The NER helps to detect relevant entities from user's message. Our proprietary Named Entity Recognition (NER) engine is designed ground up for chat bots and is 1st in the world to be open sourced for anyone to use. You can find the open source repository here.
Once we figure out the domain and entities, we figure out the intent of the conversation and the chat goes through a conversation tree to figure out the right response.
Using millions of data points built up through previous conversations and third party data, the machine learns to respond back to certain queries it may have never seen before and send an accurate predictable response.
The ML orchestration layer takes input from all the different algorithms and picks the right response to continue the conversation. This layer helps bring everything together and makes the system more scalable. We can add and remove any algorithms as long as they follow the same protocol. This layer also helps send data to our date lake which powers our analytics engine.
Haptik’s agent dashboard is built for active monitoring and intervention by human agents. Agents can pick up chats that aren’t addressed by Bots and course correct to ensure great user experience.The intuitive dashboard not only auto allocates conversation to available agents, it also measures agent productivity, efficiency and has in-built escalation matrix.
The Haptik engine collects data in realtime and delivers classical analytics on to of it!On Top of the data collected we have an extendible UI layer for reporting that enables us to expose specific data cuts for partners, thus delivering rich insights. There is also monitoring built on top of the data which raises an alarm whenever automation numbers fall.
Today, we are open-sourcing a part of our proprietary technology powering Haptik’s apps – Chatbot NER. Chatbot developers across the globe can use Chatbot NER straight-out-of-the-box to build intelligence into their chatbots.
In this blog, we talk about how to make Android debugging easier using Stetho Library open sourced by Facebook.