The Conversational AI market is growing dramatically, training new large generative models, expanding the technology stack, and bringing more advanced products to the market. By 2030 the market size is expected to reach ~42 billion dollars. The technologies behind conversational AI products are becoming more complex as well. 

And that’s no surprise — tech giants invest vast sums of money in R&D. For example, the project OpenAI (founded by Elon Musk and other tech stars in 2015) got a $1 billion endowment and raised $1 billion more from Microsoft in the next three years.

At the same time, wide usage of the modern and advanced ConvAI, except for large generative models like GPT-3 (Generative Pre-Trained Transformer 3), is still limited. While many toy projects with GPT-3  are in place, and some chit-chat products use it and similar models, other advancements are still not as widespread. How can smaller companies benefit from technology development and implement AI assistants today?

Expectations vs. Reality, is the team behind the open-source Conversational AI stack designed to enable faster and easier chatbots and AI-assistants’ development. We surveyed startups about the expected results of Conversational AI usage in their businesses in June 2022.

We spoke to twenty startup founders and CTOs from the edtech, fintech, automation, and consulting markets. All of them mentioned the undeniable benefits of chatbots for their niches, including making businesses more seamless. 

The results show that startups expect substantial results from ConvAI, among which are an increased containment rate of customer contacts, increased call center human agents’ efficiency, and an opportunity to cut operational costs. But at the same time, almost 80% of respondents mentioned struggling with implementing and developing chatbots.

Earlier, RASA also looked at the current state of Conversational AI adoption in customer service. They stated two crucial business benefits the technology offers: the ability to automate two-way natural language conversations with customers and the ability to understand customers’ needs through analyzing conversations.

These results correlate with global research studies. Gartner says that small businesses can save money on salaries and training by replacing people with chatbots.

Key results from the investment in the chatbots include increasing the number of customer contacts through a virtual assistant and improving customer experience. As well as increasing the efficiency of sales managers and providing additional business opportunities. 

What is the Stumbling Block 

The bottleneck for adopting the modern ConvAI is that startup founders lack clarity and deep understanding of the technological opportunities. They struggle with naming the functions they would require from the chatbots, particularly NLP classifiers. They mentioned this while answering our questions. 

Half of the respondents said they would benefit from extracting intents and characteristics of human speech in dialogues with clients. Another popular request, around 45%, is to extract named entities. And 10% said they require classification of sentiment characteristics of human speech.

It shows just how vague the picture of Conversational AI founders often have. The majority (80%) do not understand all the possibilities of cutting-edge Conversational AI and chatbots for their businesses. As a result, it restricts them from the implementation of the technology.

Moreover, the more sophisticated Conversational AI platforms are becoming by adding support for multi-domain, multi-modal, and multi-industry requirements, the harder it is to use them for the average startup market participant.

So both creation of the technology and making it comprehensible are vital to the prosperous future of Conversational AI. Educating startups about Conversational AI will fill in the knowledge gaps.

Unleashing the Advantages of Conversational AI to Startups

A better future for chatbots and their usability can only happen with the dialogue between tech giants, labs, developers, and startups. 

A good example of bringing the industry together is the Alexa Prize Challenge. The opportunity to test technology in a safe space and get thousands of conversations lets teams take away unique findings.

For example, last year, one of the teams learned that 10% of users talked to the bot for more than 10 min, tried to build a relationship-type connection by asking personal questions, and even tolerated the bot’s oddities.

Amazon, for sure, is doing a great job by encouraging developers to add new skills to Alexa or use them as a foundation for their ConvAI solutions. But as its core technology is closed-source, it makes its further usage limited.

Another approach is bringing the power of open-source ConvAI solutions.

And RASA is being phenomenal, supporting the development of task-oriented chatbots with its open-source framework. But developing multiskill AI assistants with its technology remains a challenge. DeepPavlov’s team, as an academia-born project, aims to enhance open-source usage.

I make the development of complex products easier and faster for the target audiences, including small and medium-sized businesses. 

Startups can obtain huge benefits from using AI

However, a significant educational part of the work is yet to be done. To maximize the benefits of chatbots and virtual assistants, startups should know what deliverables to expect and how to develop the technology for their particular cases.

Market players should realize — the technologies’ advancement should go along with making them understandable and easy to use. 

Featured Image Credit: Photo by Tara Winstead; Pexels; Thank you!

Inner Post Images Credit: Provided by the Author; Thank you!

Daniel Kornev

A chief product officer for which is the developer behind the DeepPavlov open source Conversational AI stack for building voice assistants. The company has over 50,000 downloads and 5,000 Stars on GitHub. Also served as an advisor to the Alexa Prize team from the Moscow Institute of Physics and Technology. Before DeepPavlov, spent two years working as a consultant and interim executive with a variety of conversational AI startups. Was a Senior Product Manager at Yandex where he worked on some core feature sets for the Alice voice assistant. In 2012-2017 he founded and led an AI-driven startup Zet Universe which main product was a semi-automated dashboard for information workers like PMs. It integrated project info from different sources into a personal knowledge graph and then enabled users to visually organize this information by projects to see each project’s state in one place instead of looking it up in all the apps project information was coming from. Earlier, he was a technical program manager at Google and a program manager and dev evangelist at Microsoft. Has an M.S. in computer science and has done extensive research in human-computer interaction.