Rev.ai is an automated speech recognition offering by Rev.com for developers that can make audio and video content searchable and accessible. They built rev.ai from millions of hours of human-transcribed content from Rev.com's om-demand audio transaction services. Rev has two primary APIs. One is for rev.ai's speech to text API, which performs automated speech recognition on input audio / video. The other is Rev.com's transcription and caption API, which allows integration with their human transcription and captioning service trusted by over 100,000 customers including Visa, Salesforce, and Disney. In addition to exposing these APIs to customers, Rev also uses these APIs internally for their mobile apps including Rev Voice Recorder and Call Recorder.
As VP of Engineering for rev.ai, Dan Kokotov oversees the team that develops these APIs. He needed a solution to be able to monitor and analyze that API traffic. They already had internal logging of API calls to Kibana, an on-premises general-purpose logging tool. However, they couldn't get the advanced filtering and high-level analytics they were looking for from logs alone. They were also looking for a solution to aid customer success in debugging specific customer issues and something to get automatic warnings when unusual patterns happen on the API. They looked at a variety of solutions including API proxies/gateways. Their main alternative was just sticking with their internal logger.
"There were really no other services we found that offer this kind of feature set, except for much more heavyweight API proxy / management services like Apigee."Dan Kokotov, VP of Engineering for rev.ai
He didn't want to go the API gateway route, as they don't really need any of the other features those services provide. Rev built a custom data collection integration on top of our C# API lib which does not sit in the path of API calls and only captures data passively (since then, Moesif released a higher level .NET SDK on top of the C# API Lib to make integration even easier). The Rev team has been successfully using Moesif for deeper analytics that logging doesn't provide. They especially use Moesif when wanting to do a bit more digging to see if an issue exhibits certain patterns, and to proactively to explore API usage patterns understanding how their customers use their API such as time of day, geo location, etc. Moesif has reduced the time to debug new customer issues that pop up.