Published in


Introducing Live Training for Chatbots

Live Training allows conversation designers (or the person responsible for making day-to-day bot updates) to intervene in active conversations and build bot responses in real-time based on real user data. This feature allows for rapid, iterative bot development.

It is great for internal testing with teammates or clients. You can build your bot from their simulated user conversations. And you can use it post-launch, which is when most the work happens based on the actual user inputs your chatbot receives in production.

Live Training is exclusive to Pro Plan users with the Chat Widget or Landing Page, but you are free to test drive this powerful new tool during our two week free trial.

To use Live Training, you will need to deploy your chatbot using the Chat Widget or the Landing Page. If you already have either of these deployments, you will need to redeploy with the updated Live Training settings.

You can find Live Training in the left-hand navigation for your bots.

Any conversation on the new deployment will show up in the Live Training Page.

Select an active conversation to view the existing conversation between your bot and the end user (teting1 and Rusty Robot Devil, in this instance).

All of the active conversations will show up on the left-hand side

If your bot isn’t responding well and you need to build responses to this subject, you can then take over. Either toggle the bot to paused, or just enter in a response to pause the conversation.

Light purple are bot or automated responses; dark purple are live agent responses

This will trigger an automated message informing your end-user that a human is taking over. This message is configurable in Live Training Settings, so you can write it in any language. After updating your bot with more satisfactory replies, you can pass the message thread back to the bot, or continue chatting as a human agent.

All of the responses that you entered will be saved to pand_livetraining in your AIML files. Each category contains that to make sure that the responses aren’t overly broad if you publish right away (for instance, if the user said “Yes”).

You can publish right away, or make simple changes to the patterns and then publish. Your bot just grew from what your users were actually talking about!

Upgrade to a Pro plan to use Live Training for your bot today!



The leading platform for building and deploying chatbots.

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