Forget about the chatbots, build workbots instead!
What it does
@Elio is an Amazon powered conversational bot that is able to assimilate natural language inputs using tur.ai’s deep learning knowledge base which was developed by analyzing recordings of 10 000+ hours of team collaboration and instruction that your DevOps has had to execute in the past. Currently @Elio knows how to:
-Set up Jenkins and Wordpress- -Add Amazon ARN role to enable all configured AWS services for users- -Show S3 buckets- -Grant SSH access to servers-
and lot more to come soon
How we built it
We’ve developed tur.ai’s deep learning knowledge base through 10 000+ hours of recording and we have then combined that with Amazon Lex and Lambda to create a workbot that is fun to engage with but also gets the job done. Elio’s ability to do many functions at once can save your team time which will save money in the end. Added to this Elio’s accuracy is exceptional. This means you don’t have to navigate through human error.
Challenges we ran into
Main challenge was teaching @Elio to be multi-tasking. We’ve used a combination of TUR.ai, (a 3rd party service developed by WeAreBrain), Lex, and Lambda- functionality, to allow Elio to queue requests and action them concurrently.
Accomplishments that we're proud of
We really proud we were able to make @Elio truly conversational bot and provided him with a few intents that allow him to continue a conversation on non-work related subjects. We’ve even included some small talk prompts and a bad joke here and there.
What we learned
Amazon Lex is a great technology, we've been searching for it for quite a long time as it finally enables potential of Tur.ai. As developers we liked diversity, when you have straight integration via hook to i.e. Slack together with Lex REST API. For our challenge of making @Elio conversational we used both, to tie elements together and this way humanize conversation and tap onto Tur.ai knowledge base. Where we would like to see improvement with Lex's Error handling, specifically let custom Lambda function to catch errors and let developers decide how to process them. In our case that would improve communication between @Elio and team in channel, where we could keep @Elio on a standby most of the time and only activate when he is mentioned.
What's next for @Elio
As of August 2017 @Elio will become publicly available at Slack's App Directory. There is an extensive roadmap of skills and integrations we will be teaching @Elio to, you can check it here