The Future of Chatbots and Voice Assistants

As 2020 comes to a close, leading experts in Chatbots and Voice Assistants weigh in on the biggest trends we have seen as well as predictions for next year. 

The experts include: 

• Alex Weidauer, CEO/Co-founder, Rasa
• Braden Ream, CEO/Co-founder, Voiceflow
• Caitlin Gutekunst, Sr Director, Marketing & Development, Creativity, Inc
• Casey Phillips, Product Lead, Conversational AI, ADP
• Connor Cirillo, Head of Conversational Engagement, HubSpot
• Dmitry Gritsenko, CEO, Master of Code
• Elaine Lee, Principal Product Designer, Twilio
• Kat Zdan, Conversational Designer, Google
• Mihai Antonescu, Product Manager, Mercedes-Benz R&D North America
• Milkana Brace, CEO/Founder, Jargon
• Roger Kibbe, Sr Developer Evangelist, Samsung/Viv Labs
• Upkar Lidder, Software Engineer, IBM

Trends of 2020

Increased enterprise demand for chatbots and voice assistants

This past year has been crazy to say the least. The Covid-19 Pandemic has had a deep impact on all our lives, including how we interact with each other and businesses. 

One of the biggest trends this year was the increased interest in automated, AI chatbots for customer service. Given shelter in place rules, social distancing, and cost cutting measures due to the pandemic, there was an increased demand for automated solutions. 

As Milkana Brace of Jargon states, there was an “explosion” of chatbot usage and adoption due to the pandemic.

Early on, some sectors were hit even harder, like the travel industry where consumers could no longer travel, or did not want to, and had increased questions around cancellation rules and refunds and/or whether flights, hotels, or tours were still even operating. The need for automated solutions to handle increased demand and provide good customer service was clear. 

Similarly, in the financial services industry, banks wanted to provide high-quality, customer service especially at a time when customers were unable to enter a branch in person. As Casey Phillips of ADP points out, virtually every major financial institution has implemented some form of basic chatbot. 

The use cases were not just in customer service, however. Connor Cirillo of HubSpot saw an increase in adoption of chatbots for marketing purposes to capture and qualify leads. As he states, “it’s a great place to start,” and there is more innovation to come. 

Dmitry Gritsenko of Master of Code saw an increase in health care related chatbots as well as traditional offline businesses moving online via a chatbot experience. 

Not only have text-based chatbots taken off during Covid-19, so have voice assistants. Caitlin Gutekunst of Creativity Inc, saw an increase in interactions with voice-assistants as families adopted new technologies and habits around their newly disrupted routines. 

Underlying the increased demand for chatbots and voice assistants, was the movement of automated-AI initiatives out of experimentation and innovation teams, to the business units. As Alex Weidauer of Rasa points out, enterprises started treating AI as a product instead of just a project. 

The move to multimodal experiences

A common trend in chatbots and voice assistants is to go where the users are - to be available on the interfaces with which the customers want to interact. 

Users are not just using the web, but mobile, voice, and touch interfaces too. 

Even the “voice first” initiative is not “voice-only,” but rather keeping voice in mind when building a solution. 

Multi-modal experiences were one of the biggest trends Kat Zdan of Google saw this past year.

At IBM, Upkar Lidder saw a trend in the financial sector for enterprises to integrate across different channels, to more seamlessly handle the hand-off from one channel to another. While there have been improvements, he sees there is still a gap between starting a conversation on the phone and continuing via another channel like SMS. There is an opportunity for enterprises to implement a more holistic approach. 

Related to cross-channel, multimodal interfaces is the context. Where the user is, and what they are doing, is important. 

As Mihai Antonescu of Mercedes indicates, usage of voice products depends on users’ physical context. For example, as he points out, recipe voice skills will most likely be on smart displays in kitchens where the users are when they need them. Similarly, if a user is driving, they are probably more interested in using voice interfaces for car controls or point of interest search, while getting from point A to B. 

Mihai adds that traditional perspectives on use cases may have more traction with a twist that takes context into account. For example, while a user in a car may use a home automation voice skill to turn on or off lights when they are nearby their house, they may be more interested in checking the status of the lights as users tend to worry more about the status - did they leave the lights on? 

The rise of independent voice assistants

An interesting trend in 2020 was the rise of independent voice assistants, outside of the traditional Alexa and Google interfaces. 

As Roger Kibbe of Viv Labs mentions, the growth of independent voice assistants, whether internal or based on platforms like Houndify, was one of the things that really stood out in 2020. 

Braden Ream of Voiceflow saw a similar increase in adoption of independent voice assistants. 

In addition to the more traditional voice experience like Alexa and Google Assistant, enterprises have been experimenting with custom voice applications to provide better experiences than traditional IVR. 

Improvements in AI and NLP

As the chatbot and voice assistant space continues to take off, a big trend this past year was in the improvements in the underlying AI and NLP.  

As Alex points out, the excitement in NLP research was huge! OpenAI’s GPT-3, in particular, captured a lot of people’s interest this year. While it is exciting, there is still more work to be done. 

Elaine Lee of Twilio adds that there has been an increased desire to create unbiased, inclusive AI-assistants - starting with the training data. 

The underlying platforms continue to evolve, and leading enterprises have been moving beyond the basic use cases to improve the overall experiences. Companies like Intuit and GoDaddy have been working to make their automated, assistant experiences more proactive to anticipate user issues and respond, instead of being completely reactive. 

Predictions for 2021

Further increase in enterprise adoption

A common prediction for the coming year is the increased adoption of chatbots and voice assistants by the enterprise. 

Alex predicts the trend for enterprises to view conversational AI as a product, rather than a project, will accelerate. Enterprises will employ interdisciplinary teams to build, run, and iterate their digital assistants. 

As Connor puts it, savvy companies will embrace chatbots, as they can provide a great customer experience and help with upgrade and retention rates. 

Both Roger and Braden see a continued growth in independent voice assistants across the enterprise. Roger further predicts that companies who do not have a voice presence by the end of 2021, will be left behind as their competitors use voice as a strategic advantage. 

Caitlin sees a gradual shift from voice as a marketing channel to an interface for interacting with technology and digital ecosystems. 

The increased adoption prediction is not just limited to the enterprise. As Caitlin points out, the increased usage of voice assistants during shelter-in-place, may lead to consumers being more comfortable talking to voice assistants outside the home, and thus lead to new use cases and experiences. For example, she mentions new Alexa enabled toys. 

Richer, proactive, and predictive experiences

Not only is the interest in chatbots and voice assistants anticipated to increase, but so is the quality of the experiences. 

As Casey points out, enterprises have a lot of data they can leverage to offer even better experiences. For example, he would like to see chatbots that are more proactive - anticipating and predicting users’ needs.

On a similar note, Elaine would also like to see more intuition based AI assistants. She envisions assistants that can communicate with non-verbal and non-conversational cues like sounds or actions, to anticipate and deliver what users want. 

As an advocate of human-centered design, Kat is excited to see new experiences that solve real problems for the user. For example, given the pandemic, loneliness is a bigger issue than ever - Kat is excited to see how chatbots and voice assistants can provide meaningful comfort and care during these times. 

Mihai predicts that multimodal interfaces featuring both voice and touch will take over as the best way to interact with digital devices. He sees more voice assistants being paired with apps in smartphones, smart displays, and in-car infotainment systems. 

As these experiences get better, and more brands adopt experiences, Caitlin predicts there will be even more focus on data privacy.

Advancements in NLP / AI

The underlying AI and NLP platforms will continue to improve. 

Upkar predicts advancements in NLP solutions that enable building generic chatbots based on large data models. He envisions improvements that enable using AI and ML to build the chatbots, instead of manually training Intents.  

At the same time, Dmitry envisions conversational applications moving beyond simply AI to include emotional intelligence. 

Despite the pandemic, we are living in exciting times and already starting to see signs of the predictions. It will be interesting to see how the underlying infrastructure continues to evolve and how enterprises leverage the technologies to develop new user experiences. 

Hopefully we all get back to a healthier world soon in 2021!

In their own words...

Below are the trends and predictions from the experts. 

Alex Weidauer, CEO/Co-founder, Rasa

2020 Trends

1) COVID-19 has accelerated the need for improved contact center tooling and conversational AI: Legacy contact center stacks reached their limits with increased digital communication and agents working from home.

2) From project to product mindset: Many enterprises start to see conversational AI as a product instead of a project, bringing in existing software engineering workflows from web/mobile like CI/CD. It becomes increasingly important to learn from real user messages (see also: https://blog.rasa.com/conversation-driven-development-a-better-approach-to-building-ai-assistants/)

3) Excitement about NLP research is huge: GPT-3 captured everyone's attention this year. But it's not all great: https://blog.rasa.com/gpt-3-careful-first-impressions/

2021 Predictions

The trend from 2020 to see conversational AI as a product will accelerate. As the initiatives (especially for customer-facing use cases) become larger and more mission critical, interdisciplinary teams will build & run assistants, and improve them using real user messages (conversation-driven development). This will lead to a better end user experience and higher levels of automation.

Braden Ream, CEO/Co-founder, Voiceflow

2020 Trends

The rise of independent voice assistants.

2021 Predictions

We're going to continue seeing independent voice assistants grow in adoption across the Fortune 500 for both internal automation and external customer-facing use cases.

Caitlin Gutekunst, Sr Director, Marketing & Development, Creativity, Inc

2020 Trends

The COVID-19 pandemic has accelerated interaction with voice assistants, as families try out new technology habits and develop new daily use cases around their recently disrupted routines. As consumers shift out of shelter-in-place, they will be more comfortable talking to voice assistants in "unusual" locations: at retail (no touching, please!), in the car, in hearables/wearables and even in classrooms and workplaces.

From this increased usage will follow better voice models and even more voice-enabled devices to hit the market. An example of what this landscape might look like could be found in the first voice-controlled toys that recently launched, which Creativity helped to develop: Fuzzible Friends, a $19.99 Alexa-connected plush toy from Jazwares, and KidKraft's Amazon Alexa Enabled 2-in-1 Kitchen & Market - heralding in new ways for kids to play that are immersive, highly interactive, and offer great play value.

2021 Predictions

1) There will be more focus on data privacy, with brands wanting a direct relationship with their consumer and proprietary data ownership, and consumers wanting more ways to control how their data is used.  

2) There will be more focus on voice as an interface for interacting with technology and digital ecosystems, and a gradual shift away from voice as a marketing channel; monetization of voice apps has not manifested as expected based on the mobile market, but is instead a way to deepen customer relations and increase efficiency.

Casey Phillips, Product Lead, Conversational AI, ADP

2020 Trends

Actionable and transactional conversations were the biggest trend I noticed in 2020. Virtually every major financial institution has implemented some form of basic chatbot that performs simple actions like displaying your recent transaction history or showing your account balance. Hopefully users truly value performing these actions within the conversation versus the traditional UI, otherwise, the value proposition for users is really going to miss the mark in the next few years.

2021 Predictions

I hope to see a lot more proactivity and predictions in the chatbots. There is so, so, so much data that virtually every company has on their users that could be put to great use in these scenarios. There are so many instances in which the data likely exists to make a fairly solid prediction in regards to what users need. I'd love to start seeing more of this in action!

Connor Cirillo, Head of Conversational Engagement, HubSpot

2020 Trends

I saw way more companies start to be comfortable with and embrace conversational marketing this year. Most of what I've seen has been using onsite chat to collect and qualify leads. It's a great place to start and shows that there's a lot of innovation still to come.

2021 Predictions

Savvy companies (particularly in SaaS) will embrace in-app chat. By definition, a user that's actively using your product is invested in being successful. Offering them a way to connect with someone about setup or usage is a great customer experience — and great for upgrade/retention rates, too.

Dmitry Gritsenko, CEO, Master of Code

2020 Trends

- Rise of the medical and healthcare bots (because of COVID and high demand for health and medical care). 

- Chatbots for offline businesses going online (a lot of business were forced to go online and the chatbots were the easiest and fastest solution for them)

- Teams collaboration automation (due to work from home transition)

Overall accelerated pace of innovation: less bureaucracy, more readiness to launch betas

2021 Predictions

- Most messenger platforms will get native payments support, that will make Conversation Commerce more convenient

- Conversational applications will leverage Artificial Emotional Intelligence vs simply AI 

- Mitigation of the "AI Biases" in the Conversational Applications

Elaine Lee, Principal Product Designer, Twilio

2020 Trends

A desire to create unbiased and inclusive AI-powered assistants, starting with the training data.

2021 Predictions

I would like to see less conversation and more intuition from AI-powered assistants. They will communicate with non-verbal and non-conversational cues, like sounds or immediate action. Individuals can opt in to allow assistants to anticipate and deliver what they want based on learned behaviors.

Kat Zdan, Conversational Designer, Google

2020 Trends

Multi-modal experiences 

2021 Predictions

I am a huge advocate of human-centered design. I always love to see designs that focus on solving real problems for the user. 

With the pandemic, loneliness is a bigger issue than ever. I’m excited to see how chatbots and voice assistants can provide meaningful comfort and care to people during these extraordinary times.

Mihai Antonescu, Product Manager, Mercedes-Benz R&D North America

2020 Trends

Customers naturally adopt context-specific skills in the space they're relevant for! Today's voice assistants provide a large variety of functionality and have made certain fulfillments more accessible and approachable. With so many options when it comes to devices that have voice assistants we would imagine that usage and take rates would be similar to the ones of apps from smartphones and yet usage on voice assistants varies quite a lot. The trend that I’ve been noticing this year is that developers primarily focused on context-specific skills in the space they’re relevant for, since customers naturally adopt those features in that physical context. For example asking for recipes in the kitchen on a smart display or performing a POI (Point of Interest) Search using voice while in the car. I believe that these high usage rates for specific skills in a space are because of habits that we formed after the voice assistant solved problems for us. This created an internal trigger that brings us, the users, back asking for more. Given that we spent more time at home this year we had more opportunities to engage with our voice assistants thus allowing the habits to set in.

2021 Predictions

Multimodal UI, i.e. Voice and Touch, is the next big thing! It will be largely driven by voice in apps! Voice UI brings many benefits over traditional touch interfaces such as:

• We already know how to speak a language and we communicate on a regular basis with other humans
• It can control any function from the top level. No need to select a specific app and find a graphical action button.

One big disadvantage for voice UI is that it lacks DISCOVERABILITY since without a graphical interface there are no signifiers that can help customers learn about the assistants functionality. A graphical menu can communicate affordances much easier and give an overview of capabilities to the user in a much shorter time.

I predict that in 2021 we will see more voice assistants being paired with traditional GUIs within apps in smartphones, smart displays and in-car infotainment systems. While the latter two platforms have been pioneering multimodality between voice and touchscreen for the last couple of years, the mass adoption and true acceptance by the customers will happen when the smartphone market fully embraces this next generation of interaction. This will be a first step towards true multimodality where a touchscreen device sees your gestures, listens and talks thus enabling the most natural UX.

Milkana Brace, CEO/Founder, Jargon

2020 Trends

Explosion of chatbot usage and adoption fueled by the pandemic

2021 Predictions

Continued growth in custom assistants (chat and voice)

Roger Kibbe, Sr Developer Evangelist, Samsung/Viv Labs

2020 Trends

1) Independent voice assistants. While the big voice assistants saw some incredible gains in 2020, what struck me the most about 2020 is the growth of independent voice assistants. Some of these were developed internally and some were based upon platforms like Houndify 2) Growth of multimodal - more and more consumers are purchasing or using voice on multimodal devices whether they be smart speakers, phones, watches, appliances, TVs, hearables, wearables and more.

2021 Predictions

1) Continued growth of independent voice assistants 2) Standards like those being worked on by OVN (Open Voice Network) will start to emerge and at a minimum discussions on supporting them will occur 3) Companies that don't have a voice presence by the end of 2021 will be left behind and their competitors with a voice presence use this for a strategic business advantage

Upkar Lidder, Software Engineer, IBM

2020 Trends

The big trend with our enterprise clients is integration between the different channels of communications with their customers. I am seeing this in the financial sector. There is still a gap between calling into a business and then continuing the conversation via sms or social media messages. Clients are still building single chatbots and missing the wholistic picture.

2021 Predictions

I predict advances in NLP solutions where we can build more generic chatbots using very large models. There is a large interest in open source models by companies like huggingface and I expect this trend to increase next year. Another trend is to use AI and ML to build chatbots for us instead of us manually teaching intents etc.