Sometimes your bots will present a specific set of solutions to your customers. Your customers may need help with different problems or be unable to articulate what they need help with. They like to be heard and prefer more human and personalized conversations. Fortunately, the chatbots in Freshchat are powered by intent detection. Intent detection enables your chatbots to understand what the customer needs help with and direct them to the correct Flow, Q&A, FAQ, or even conduct small talk with them.
Note: Please refer to this solution article to learn more about intents, utterances, and customer queries — which is necessary to understand how to map intents.
Triggering intent detection in your bot flows
- Login to your Freshchat account instance as an Admin > Navigate to Bots from the sidebar > Select an existing bot, create a new bot, or use a template.
- If you're using an existing bot, make sure you’re setting up Intents for the latest version of your bot.
- Navigate to the Flow to which you need to add intents. When your bot detects this intent, it will redirect the customer to this Flow.
- Click the Map Intents icon next to the title to open the Intent Mapping screen.
- To start afresh, click on New Intent. You can add a new intent by typing the Intent name in the Search Intent box. Based on the search text, similar intents appear as a list, and you select them to add new utterances. You can click the Create message to add this new intent if there are no similar intents.
- Give a name to the Intent and add at least five utterances to map your intent to the selected Flow. Follow the best practices to ensure the bots are trained well for any utterances from the customer.
- Once you have configured the intent and the variants, you can map it to the Flow by clicking Save & Map to Flow.
- Once you've mapped an Intent to a flow, you can either make it inactive or delete the intent. You can access these options by clicking on the Intent name in the flow
- Use the Add utterances button to add multiple new utterances to the same intent.
- The bot builder also provides validation for the utterances you add. You will be notified if you add the same utterance more than once or if an utterance is better suited to another existing intent.
- You can perform quick actions to Move to the detected intent or Ignore it to continue mapping your current intent.
Using the intent library to map intents to your bot flows
- Sometimes, you may not have a ready list of customer queries you want to map as utterances to identify intents for a bot flow. You can use the intent library from the sidebar to help you get started quickly.
- The right navigation bar gives you templates of intents from various industries. You can filter, explore and edit these templates per your requirements or create a new intent tailored to your needs.
You can set up intents across all the languages you want to support your customers.
- Select the language drop-down and pick a language from the list of configured languages to map utterances.
- Even if your customers ask your bot questions in a different language, the bot will translate, compare, and find the closest match among a flow, Q&A, FAQ, or Small Talk.
- Once saved, training your bot with the added intents is essential. Click on Train Intents to train your bots to learn from the new utterances. Your bot will also be trained when you Publish the bot flow.
Detecting intents in a bot flow
Once you've finished setting up many questions to enable your bot to recognize the intents for each question, you can start using Natural Language in the bot flows.
- If you're deploying your bot on the Freshchat widget you can detect the customers' intents in two ways:
If you're offering multiple choices for your customer, you can detect customer intents using a quick action for them to type their response freely (instead of picking one of the options)
If you're letting your customers respond directly to your question, you can trigger the Detect Natural Language action to detect their intent.
- For customers using third-party channels such as WhatsApp, you can use the predefined Accept natural language input checkbox to detect intent. This will be useful as customers can type their queries on third-party channels like WhatsApp — even if other options like buttons are available.