For some questions, you need the advice of an expert.
But for many technical questions, it can be hard to find the right expert or the precise answer.
Oftentimes, developers search Google, comb online discussion forums or try to get help from whatever support service is offered by the development software they're using, said Robin Purohit, CEO and co-founder of Peritus.ai Inc. But such efforts typically yield far more misses than hits. Even on the top technical forums, only about a third of developer questions get answered at all and only 20 percent are answered within the first day.
“It becomes a multichannel hunting experience looking for the best advice you can get," Purohit said. "It becomes very time consuming, and you have to evaluate different recommendations you get from people."
Purohit thinks he has the answer to this problem. Peritus, his Palo Alto-based startup, has developed an artificial intelligence-based service that's designed to help developers of cloud applications get the answers they need in a timely fashion.
Peritus, which means expert in Latin, trained its software on more than 27 million questions and answers that were published on product forums, GitHub and in conversations on Stack Overflow. Its system gives higher rankings to answers given by qualified community experts, so that developers can easily find the most credible responses to their questions.
The company debuted its Peritus Assistant earlier this year as a Chrome extension, allowing developers to find the best answers to their questions as they were surfing the web. On Thursday, the company launched Peritus Assistant for Slack, so now developers can find the answers to their queries through the communications app.
Data is important in artificial intelligence work
Santhosh Srinivasan cofounded Peritus in 2017. Purohit joined the team after meeting Srinivasan at The Hive, an incubator that focuses on startups working on artificial intelligence-based products. In 2019, Purohit sold his prior startup, Griddable.io to Salesforce.com Inc.
"I knew I wanted to work at an AI company after I sold my last one," Purohit said. "When you’re trying to do something new and the industry’s moving in a new direction, an incubator is invaluable."
Initially, Purohit and Srinivasan focused on areas where there wasn't a lot of data available. But then they realized that there's troves of data for developers on online forums and other such services.
The biggest mistake people make when starting an artificial intelligence company is to underestimate the amount of data they'll need to train their systems, Purohit said.
"Curating data that is of high quality and has a lot of signals in it to make predictions and inferences is a big part of a problem," Purohit said. "You have to pick a market that has enough data available for these machine learning systems to produce something."
Peritus now has 19 employees and has raised a total of $7.8 million in funding. That amount includes $3.4 million in seed funding it announced Thursday that came from investors including First Ray Ventures and Benhamou Global Ventures.
The company is looking to take a slice of the $36 billion customer self-service market. But because Peritus is creating a new kind of service, it's not easy to estimate the potential size of its particular piece of that market , Purohit said.
"The more interesting stat is that there’s approximately 26 million developers doing enterprise development," he said. He continued: "That’s the market that we're focused on long-term."
