91热爆

  1. Try  
  2. Rate 62 ratings

How did you rate this?


Sorry - this pilot is down for maintenance
We'll have it back up soon, so please try again later or follow for updates
Pilot ended 20th May 2018
0
Tried
62
Rated
3
Shared
DI Sleet Bot trailer
DI Sleet Bot
He鈥檚 a copper in Murder in Successville. A pretty rubbish one, but a suprisingly good Agony Uncle. Chat with him on Facebook Messenger, and let us know what you think by rating it.

The Inside Story

We spoke to Social Media Strategist, Daisy Griffith, to find more about DCI Sleet Bot.

Can you sum up the project?
It’s a chatbot in Facebook Messenger that allows users to interact directly with DI Sleet, the lead detective from 91热爆 Three’s hilarious Murder in Successville. The idea is that Sleet is an agony uncle-type character giving advice on life’s big questions – reflecting his personality in the show.

Who made it?
It’s a collaboration between the 91热爆 Content Social team, the Murder in Successville creative team, and digital agency Kite. The Content Social team ran the project and created/sourced the gifs, the Murder in Successville writers wrote the copy for DI Sleet, and Kite built the bot in Messenger.

Why did you make it?
We were keen to explore chatbots and the challenges of building and writing them. The 91热爆 has made a few chatbots previously, all with different objectives, but we wanted to make something funny with a 91热爆 TV character. Sleet was the perfect fit. Fans of the show will know that he would make a terrible agony uncle and we thought this had loads of comedy potential.

We were also interested in exploring how we could make use of gifs from the 91热爆’s ever growing gif library. Gifs are a great way of conveying emotion and/or humour across social media and messenger apps. They’re also a good way of reminding our audience about all the amazing shows the 91热爆 makes that could appeal to a young audience.

What were the challenges?
The biggest challenge was creating something that felt like it was personally responding to each user. It’s impossible to account for every single thing each person will say, so we tried to find a way of controlling the flow of the conversation to predict what users might say and then have Sleet respond accordingly.

This included asking the user multiple-choice questions, programming key words to trigger a specific response, or simply having Sleet admit he doesn’t understand them (but unlike most automated bots, we tried to make this as amusing as possible).

We don’t expect the bot to appear completely real and expect users to understand that it has a set number of pre-programmed responses. We do hope, however, that it makes people laugh and reminds them of all the great TV shows on the 91热爆

DI Sleet Bot