To summarize the issue, I have an application which has integration with dialogflow. In dialog flow I have configured the intent. For each intent I have created a webhook.
Wanted to understand if I have to return data from a data base along with the fullfilment for the intent what will be the best way to do?
Related
I have a free SNOW developers instance ,and i want to be able to raise an incident from Google DialogFlow. I have the user name and reason for the call transcribed into text and need to pass it to SNOW to create the incident. I have already created an integration where i can query the status of a SNOW incident from Google.
Here you go, you can do this with fulfillment.
Enable fulfillment on that intent.
Write it in NodeJS or JAVA or python.
Raise the ticket from your fulfillment (use service now API make http call from your server)
Once you get the response, send the ticket number back to google assistant, reply to the user.
This the way you can implement this. Directly from dialogflow you can not integrate any other 3rd party API.
I am creating an app using Dialogflow and Actions on Google Node.js library.
The logic I am trying to implement is as follows:
User says: Initiate chat.
That user request triggers my webhook which then supplies the agent response.
User replies back to the agent - this is the tricky part because the user reply can be anything, thus the wrong Intent or the Fallback Intent can be triggered.
Is there a way to allow for the unpredictability of a user's response and to keep conversation going between user and agent? In another words, always trigger the same Intent (Chat Intent) regardless of what the user says?
In Dialogflow, you can configure the Default Fallback Intent to connect to web fulfillment. In your fulfillment, you can then route the user's query to your webhook and process a response however you'd want.
I have created a Google Home app in DialogFlow and have since had the app published by Google. I am assuming that I can change the fulfillment endpoint address in Dialogflow and this won't affect my live app - my assumption being that a packaged version of the app together with the endpoint at submission time is what is running on the Google Assistant.
Does anybody know if this assumption is correct?
No, this assumption is not correct.
By default, changing the fulfillment URL in your Dialogflow project will change it for the live Action.
Although there is some information that is generated when you release an Action built with Dialogflow (the Intents and Entities, for example, get turned into entries for the Assistant), the fulfillment URL is not one of them. The Assistant sends it to a special URL on Dialogflow's servers, and Dialogflow then sends it to the URL you've specified.
However, you can look into the Dialogflow Beta feature that supports versions and environments to see if that will do what you want.
I have a Dialogflow agent which is integrated in an app and a web version. I want to integrate this chatbot with Chatbase to get analysis metrics of the bot. However, as of chatbase documentation we cannot integrate Dialogflow hosted bots with Chatbase but if I have a server in the middle of my app and chatbot which is saving all the data, can I integrate chatbase with it?
The typical way to integrate Chatbase is to have it called as part of fulfillment. This requires you to set each Intent to use fulfillment, but you can still set the reply as part of the Intent, if you wish. In your webhook, you would call Chatbase to record the event, but not return any response in the webhook, and Dialogflow will use the Response set in the UI.
Chatbase reports are built upon user messages and intents, however there is value in sending us the bot messages to provide context to the Transcripts feature. You can create a server to server integration to send us the JSON payloads in real time, or in batches of up to 100.
I want to train my app using API. I am using my own UI to validate the intents and just want to validate one by one for particular intent. To get intent I am using this api https://api.dialogflow.com/v1/query?v=20150910 but don't know how to train any particular intent for any sample.
There is no API for Dialogflow's training feature but you can accomplish the same result by looking at your logs and using Dialogflow's update intent API and passing in new (or removing) training phrases