Dialogflow and Appian to make conversation-driven processes

Procensol > Procensol Labs > Dialogflow and Appian to make conversation-driven processes

Brief summary

Chatbots are extremely difficult to develop, typically they require extensive training and exception handling. Through Dialogflow, the neuro-linguistic programming is offloaded to Google, meaning we can focus on Appian development instead.

Dialogflow is a SAAS suite of ‘human-computer interaction’ technologies, which performs complex parsing of language (written/spoken), including intent detection (requests for specific responses).

To resolve an intent, an outbound webhook can be invoked, relaying this information to an external integrated service. In our case, Appian API’s can then initiate processes to modify and/or return data.

Share:
Print
+100%-

Google’s natural language processing coupled with Appian’s BPM make building and deploying Chatbots simple

Despite chatbots being largely stigmatised, as most users associate them with underperforming, dysfunctional tech-support webapps from the 00’s, modern cloud-based natural language processors are much more competent and interpretive, with more to offer than generic, prepared textual responses.

Source: Matt Brennan

Author Image

Matthew Brennan

Consultant

About the Author:From Lennox Head. Love Tech/AI but honestly know more about Art (music/cinema), Gamer, Tennis player, Futurist. Traveled a lot. Studied Master of IT at QUT (riddled with student debt)

Author Image

As a consultant, I often come across clients requesting for help-sections or search engines to be embedded within their systems, typically segregated into relatively small components. For instance, Screen A may provide contextual information about the current weather, whilst Screen B presents users with traffic updates. Separating this functionality entails much development being needlessly redone.

The conventional approach to fulfilling this requirement would be to build the respective pieces of functionality in containers, when a more convenient UX would be a consolidated interface into which the user could request either traffic or weather information independently.

Despite chatbots being largely stigmatised, as most users associate them with underperforming, dysfunctional tech-support webapps from the 00’s, modern cloud-based natural language processors are much more competent and interpretive, with more to offer than generic, prepared textual responses.

I recently created an Appian-embeddable Chatbot, capable of return any requested data regarding any entities within a given system, such as user details, weather, traffic and other basic information from a single interface.

The engine for this chatbot is Google’s Dialogflow, a SAAS suite of ‘human-computer interaction’ technologies, which performs complex parsing of language (written/spoken), including intent detection (requests for specific responses).

Dialogflow also has a relatively intuitive configuration panel (compared to alternatives such as Amazon Lex), default responses, flexible interpretations (of borderline unintelligible queries), one-click integrations (with platforms such as Slack and Facebook) and entity bindings which allow context to be preserved between interactions, all of which enhance our bot OOTB (as opposed to writing all logic in-house). This chatbot can be easily repurposed for any new system, requiring only basic provisioning, endpoint configuration and training, as through Dialogflow, the NLP is offloaded to google, meaning we can instead focus on Appian development

Author image

Matthew Brennan

Consultant

About the Author:From Lennox Head. Love Tech/AI but honestly know more about Art (music/cinema), Gamer, Tennis player, Futurist. Traveled a lot. Studied Master of IT at QUT (riddled with student debt)

Author Image