Release notes: Tangowork v2
Web management console
Edit system messages: System messages, like prompts or confirmations, are now accessed and editable from a new “Settings” menu item. This keeps the main messages interface cleaner and focused on content that changes more frequently.
Organization toggle: Do you have more than one bot? If so, you’ll like the new context-switching menu that lets you toggle between the organizations you have access to.
Preferences: Organization preferences — botPrefs, wwwPrefs, and apiPrefs — can now be read and written via the /api/orgs method.
API scoping: Every API token is now restricted to particular methods on particular calls to particular organizations within particular environments. That’s pretty particular. And, therefore, very secure.
Message configuration: For any given intent, you can now choose whether or not to respond with a message’s title, body, and image. Set any of the following to true or false: intent.sendMessageTitle, intent.sendMessageBody, and intent.sendMessageImage. You can also now choose whether or not to combine a message’s title and body in a single message, or whether to send them as separate messages. Just set intent.sendTitleBodyTogether to true or false.
Message length: The maximum length of a message is now determined by orgs.botPrefs.messageMaxChars. Longer messages are split on whitespace. The default is 320 (the maximum length for Facebook Messenger). 160 is ideal for an SMS bot.
List item separation: When Tangowork responds with a list, sometimes the list items look better in a single message, and sometimes in separate messages. Now you can choose the best option on an intent-by-intent basis by setting intent.responseSeparateMessages to either true or false.
Quotes in names: We were getting confused with Irish names like O’Leary and O’Malley, and claiming that they didn’t exist, when in fact, they did. Presumably this problem affected more than the Irish, and we fixed it.
Missing conversation context: We were losing some conversational context after prompting for required entities. We adjusted how we parse the results from LUIS, our natural language processing engine, and the problem is all fixed.