Launch in 4 weeks
The best approach to building successful, affordable virtual assistants and leapfrog the competition.
Start small, learn fast and then invest! The trick to building virtual assistants that your customers love (without costing the world) is to start small and learn from live customer interactions. To find out more about why other approaches routinely fail, download our whitepaper on ‘Why AI fails’ here. At EBI.AI we advocate building a minimum viable solution within 30 days, going live to users as soon as possible and then rapidly improving the quality and quantity of responses through supervised learning. Read on to see how we do it!
In Sprint 0, we gather information and make decisions about what a ‘Minimum Viable Assistant’ will be able to do at the end of a 30 day project.
You will gather the questions that your users commonly ask from call logs, chat logs and website FAQs, and the processes which cause your business the most pain.
The Kick-off workshop will include a UX Consultant and a Solutions Architect from EBI.AI. You should include people to represent the voice of your business and the voice of your customer.
In the first sprint, we’ll create your assistant, and configure the channels you have selected. Then we’ll start training your assistant to understand the questions you have supplied us and designing the conversation flows to get to a resolution.
In the second sprint, we’ll start adding responses. Responses will go into your assistant’s response store and we’ll build integrations to give your assistant the ability to ‘do’ things, like update your CRM, book appointments in scheduling systems, connect to a shopping basket or pull answers from a knowledge base.
After final testing and gap analysis, you have the option to go live, or lengthen the build to increase the amount your assistant can do. We’ll then work closely with you to pur your assistant live on your chosen channels. Your assistant is born and starts interacting with real users.
As your assistant interacts with real people, our highly skilled UX Consultants will review the conversations, log problems, fix issues and make improvements to existing use cases. ‘New’ use cases that we hadn’t anticipated are passed to Sprint Planning for your review. This is a VITAL process as you respond to your exact customers’ needs.
In Sprint planning, we will use analysis of the existing conversations to review the performance of the bot and ask you to prioritise the backlog of development tasks and bot training into the following sprint for us to complete. Add more channels, improve the bots understanding, perform more tasks or give it ‘memory’ by adding a Customer Data Platform.
Over time, the assistant will get better and better at responding correctly to users’ questions, and the need for AI supervision and further development will decrease. Your assistant is maturing and is left answering questions and completing tasks that your customer want, and that give you value.