Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - PME Instrument v2 (Second generation chatbot development platform)

Teaser

Advances in artificial intelligence combined with the massive adoption of chat channels and messaging apps have created a strong traction for chatbots : 80% of service companies declare they will use chatbots to interact with their clients in the coming years and it is...

Summary

Advances in artificial intelligence combined with the massive adoption of chat channels and messaging apps have created a strong traction for chatbots : 80% of service companies declare they will use chatbots to interact with their clients in the coming years and it is expected their chatbots will generate over $8 billion worldwide cost savings annually by 2022. However, the chatbot development ecosystem is still relatively immature, which means it is hard to develop chatbots that fully meet expectations from both the end user (bot performance) and the service company (bot cost and evolutivity). Most of the chatbots already in place are actually “Proof of Concept extended kind” (we call them “1st generation” chatbots) and consequently lead to 27% dissatisfaction rate from the users.

Customer support has now arisen as the most popular business case for AI chatbots. With chatbots, service companies can drastically improve the efficiency of their customer support and technical support - according to McKinsey, the cost savings is a 10x factor compared with phone support. European companies are thus facing two main challenges:
Manage to compete on quality and efficiency of customer support. Chatbots bring total availability for sales or support teams, they are able to instantly respond to an unlimited number of clients, they can find relevant answers and provide easy access to a product catalog or knowledge base…
Seize the massive financial opportunity brought by automation.

Botfuel set as its primary goal to be able to democratize and industralize second generation AI chatbots for a vast number of companies of different sizes and sectors. Botfuel identified the following challenges as being the most important hurdles to the development of customer support chatbots.
Performance: As user expectations grow chatbots need to cope with fine-grained classification of a growing number of user intents and increased complexity of interactions.
Usability: Democratization of chatbots for customer support will only happen if the cost and time to deployment can be decreased significantly which involves providing new solutions for chatbot design and chatbot training
Security and privacy: Chatbots are obviously dealing with customer data. When building their chatbot, companies are very sensitive about customer data security and especially vigilant about data leakage.

Work performed

Since the beginning of the project Botfuel’s has significantly progressed towards providing a “full-stack bot platform” solution.

Main product milestones:
Q4 2018: Launched 2 new versions of Botfuel’s framework for chatbot development (Botfuel Dialog) including additional features for advanced Dialog Management, Connectors and Webchat.
Q4 2018: Released a new version of Botfuel’s NLP engine allowing for no-training, flexible and precise Entity extraction
Q1 2019: Launched a packaged solution for Customer Support called Botfuel Answers, based on Botfuel’s technology and aimed at customer support teams. Botfuel answers is one of the first chatbot solutions that doesn’t require any training thanks to its fully pre-trained knowledge base for different industries.
Q2 2019: Launched modules for ECommerce, Hospitality and Car rental
Q2 2019: Launched new dashboard for Chatbot monitoring
Q2 2019: Launched a feature to implement pre-developed modules for generic use cases (order and reservation status, booking, faceted search, etc…)
Q3 2019: Researched and tested the use of advanced word embeddings
Q3 2019: Finalised research over multilingual bots and multilingual capabilities

Main business milestones:
Q4 2018: Pursued the development of Botfuel’s partnership program
Q2 2019: Signed contracts with several large corporates
Q2 2019: Reorganized Botfuel’s sales and marketing teams to tackle International markets and wider target user audiences for Botfuel technology

Final results

As evidenced by tests and frequent benchmarks, Botfuel’s AI technology and NLP services have kept pace with major players in the area such as Google and Microsoft. In addition, Botfuel has allocated significant resources from the project to developing “semi-automatic” and “fully-automatic” chatbot training. In recent months, Botfuel has been able to launch a new set of pre-trained models for different industries, and has opened new deployments on messaging channels, providing the following benefits:
High level of intent recognition providing a level of understanding of natural language and a customer experience that were previously missing from most chatbots
Reduced time to market as clients do not have spend time themselves designing and training the models
Large access for small and medium companies to state of the art NLP and AI chatbot technology, making it possible for small customer support teams to match the expectations of customers when the volume of requests grows at a fraction of the cost
Botfuel’s ambition remains to become a European leader in the European Virtual Customer assistant market.

Website & more info

More info: http://www.botfuel.io.