Espressive lands $ 30 million in Series B to help chatbots better


Espressive, a four-year-old startup by former ServiceNow employees, is working to create a better chatbot to reduce calls to company help desks. Today, the company announced a Series B investment of $ 30 million.

Insight partner led the round with the help of Series A main investor General Catalyst, along with Wing Venture Capital. Under today’s agreement, Insight founder and CEO Jeff Horing will join Espressive Blackboard. Today’s investment brings the company a total of $ 53 million.

Company founder and CEO Pat Calhoun said when he was at ServiceNow, he found that employees in many companies were often frustrated to find answers to basic questions. This resulted in a call to a help desk that required human intervention to answer the question.

He believed that there was a way to automate this with AI-driven chatbots and founded Espressive to develop a solution. “Our job is to help employees get instant answers to their questions or solutions or solutions to their problems so they can work again,” he said.

They do this by providing a very tightly focused NLP (Natural Language Processing) engine to understand the question and quickly find answers, while using machine learning to improve those answers over time.

“We are not trying to solve every problem NLP can solve. We are pursuing a very specific set of use cases that really speak the language of the people, so we really tuned our engine to the highest level of accuracy in the industry,” said Calhoun told TechCrunch.

He says what they did to increase accuracy is the combination of NLP with image recognition technology. “We built our NLP engine on an image recognition architecture that is really designed to be highly accurate and essentially breaks down the phrase to understand the true meaning of the phrase,” he said.

The solution offers a single instant response. If for some reason an inquiry cannot be understood, a help ticket is automatically opened and forwarded to a person for resolution. However, the latter tries to keep this to a minimum. He says that they tailor their deployment to the keywords and terminology of each customer.

So far, they have reduced the number of help desk calls to customers with an employee participation of around 85% by 40% to 60%. This shows that they are using the tool and provide the answers they need. In fact, the product instantly understands 750 million employee phrases.

The company was founded in 2016. It currently has 65 employees and 35 customers, but the new funding should increase both numbers.