Using Iris To Build a Conversational Chatbot for Twitter Direct Messages


In 2016, Twitter announced its DM API. This API allows anyone to build a chatbot and integrate it into users DMs. Along the way, there have been several updates and improvements to the API including the use of buttons, location sharing, and more. This stuff is complicated, so I’ll get into that in a future post.

Over 100+ company and brand profiles have used this API in some way or another to create a unique experience for the account. The team at Proximahas successfully built and run DM campaigns for Jack in the Box, The Game Awards, MTV, ESL, NBA, NFL, and more!

Getting Started

Step 1:
If you haven’t already signed up for access to Iris, you’ll want to request. Head to to do so.

Step 2:
If you get accepted into the beta, you’ll want to log in and find your Dialogues. Since this is probably your first time on Iris, you shouldn’t have any, to begin with. Create one by clicking New Dialogue on the left-hand side of the window.



Step 1:
Once you have created your first Dialogue, its time to start your conversational design. It sounds complicated, but the way that Iris is setup makes it very easy.

Step 2: Decide what type of nodes you want to use and where

  • Green = Quick Reply nodes

  • Greyish/blue = Conversation nodes

  • Bright blue = Regex nodes

  • Black = Decision nodes

  • Yellow = Holding nodes

  • White (You most likely won’t have these) = Advanced nodes

  • Red = Nodes that are not linked/setup correctly.

Note: Buttons and Location Nodes are coming soon.

 If you see a red node, it typically means that you don’t have a traversal connected to it

If you see a red node, it typically means that you don’t have a traversal connected to it

Step 3: Create a Welcome Message
The Welcome Message is the message that a user will be greeted with when they start a DM conversation with the account. Once you have the WM set, you can add Quick Replies.

 Welcome Message Node Settings

Welcome Message Node Settings

Step 4: Add Quick Replies
Quick Replies allow for speedy customer service. They help a user quickly find what he/she is trying to find. If you’re not a business account, that’s fine. Quick Replies can be used for pretty much anyone’s needs.


This picture might seem confusing to you, but it’s not too bad. If you visit this bot on Twitter (@IrisPowered), it will greet you with the Welcome Message and two Quick Replies (two green nodes on the left and right of the WM). Next to the Quick Reply nodes are holding nodes. They were needed because you can’t link a Quick Reply node to another Quick Reply. Linked off of the Holding node are four Quick Replies with demo information added. If you select the Learn more option through the bot, you’ll notice Buttons. Those are created with Advanced Mode which I’ll go over sometime in the new year.

Step 5: Connect your account
This is a very easy step. From the Iris homepage, press your profile button to drop down the menu > select “Accounts” > Make sure you’re logged into the account you want the bot on > Find “Add Account” and select “Twitter” > Authorize Iris

Note: It is preferred that you do not add a bot to your personal account. If you message another bot, both bots will spam each other with thousands of messages until you manually stop the dialogue.

Step 6: Open DMs
If you haven’t already opened DMs, this is when you'll want to do so. Unfortunately, Twitter is removing the prominent message button and support hours/indicators so the Support dashboard will no longer be available. Navigate to your profile settings to open them.

Step 7: Promote bot to testing
From your Dialogues Launch Pad, select the down arrow next to “Actions.” You’ll see an option titled “Promote to Testing.” Press it.

Obviously, you’ll want to connect the test version to a test account. Once you’re satisfied, go ahead and promote it Live, connect the official account and start it.


You have finished! If you followed each step, you should have a functioning DM bot on your profile. If you’d like to see this bot, visit @IrisPowered.

Have you built a bot using Iris? Let me know on Twitter @SamSchmir! I’ll add it to my Chatbots on Twitter List.