Since the email era the improvement in the communication tools has to do with speed and quality of communication, improvement in the communication hardware starting from computers to today’s smart phones, ubiquity of communication because of wireless, live video , multi-user communication like video conferencing etc.
1. Crawl Twitter Data using 30 Lines of Python Code
But here are the important ways in which today’s communication tools like email, messaging apps, social media have largely remained unchanged:
- They don’t learn from the interactions and get smarter over time no matter how long you use them. We spend a tremendous amount of thought, energy and wisdom interacting with people using these communication tools. This information just sits there forever.
- They don’t have access control capability by default, beyond detecting spam. i.e. anybody with your email address or social media handle can send you a message and it is treated the same. Note:- Gmail does mark your messages “Important” using what they called Google Magic which is kind of opaque and cannot be customized.
- Bi-directional communication only occurs when both parties are present at the same time. They don’t, by default, have an ability to respond instantaneously (except maybe an out-of-office reply).
We think having a bot-builder interface and messaging tool combination provides an opportunity to overcome the above limitations and bring these automation features to the masses. We think these automation features being default have benefits beyond mere time saving or efficiency.
With a messaging application interface, the system can intervene in your conversations, ask questions about labeling/categorizing conversations. It will try to handle those kind of queries or communication in the future. Overtime the tool becomes more and more capable.
Using the bot-builder functionality the user can create an upfront custom conversation for people reaching out to them. They can configure that conversation such that certain responses are considered “preferred responses” based on their own criteria. The system can then generate a score that helps the user prioritize the incoming messages.
The NLP/AI enables instantaneous responses by understanding what the incoming message or query is, and deciding if that query has been answered before. The back and forth involved in checking if the answer provided by the system satisfies the query and/or suggesting an alternative answer can be handled much better via a messaging interface.
If we have these capabilities we will move closer to be able to communicate with multiple-people/entities in parallel without needing to be physically present. At Spotgini, we have implemented features that give the user these capabilities in some way and move towards that goal.
We are building tools for a wide audience. Hence we decided to not use the vocabulary in current bot-building platforms which were built keeping in mind either developers or at the very least, people with technical background as the end users.
“Intents” in Conversational UX development process is what “Stylus” was to PDAs in 2007…the iPhone people correctly identified that.
Intents, Entities, Actions etc. terms which come from the field of NLP (Natural Language Processing) are part of the workflow of today’s conversational UI platforms. The need to have understanding of these concepts is an obstacle to making conversation building ubiquitous. Hence, we have decided to push ourselves to create new tools which aim to possess similar capabilities but which can enable more people without the technical background to create powerful conversations and accomplish tasks conversationally.
In the past two years we saw great advances in Deeplearning for NLP, for example, release of models like GPT-2 and BERT, which have dramatically increased the accuracy with which meaning can be extracted out of natural language. With more human-like capabilities of machines to understand language, how could we create tools that enable us to create conversations in a more human way?
Would love some feedback. If you are engaged in or interested in making conversation building ubiquitous and find this UX design + technology problem interesting lets chat.
