Olivia Chatbot

Project manager on the customer side
Ygor Silva Cezar
Head of IT
IT Provider
Year of project completion
Project timeline
May, 2022 - December, 2022
Project scope
2000 subscribers
As a B2B company, OEC does not engage directly with its costumers, but we have a huge employee base, almost 4,000 employees, and a lot of the corporate content, culture, procedures and directives is published in the corporate intranet. But almost no one uses it to access this information, they rather like to ask the HR team, legal team, IT team, etc., about the information they want to know and this takes time from these teams to provide information that is already available. So we came up with the idea to build a chatbot to be the "know it all person" in the company, Olivia, which is an acronym for Odebrecht Levando Informação Via Inteligência Artificial in portuguese, and that means Odebrecht bringing information by artificial inteligence. If anyone has any doubt about anything in the company, he or she should ask Olivia first about that.
Project Results
The usage of Olivia is growing each month and the teams that where responsible for the information that the bot is providing are reporting that they are being less contacted to give that information. With the expanse to other areas and applications, we expect Olivia to be the one stop shop for every doubt anyone in the company will refer to.

The uniqueness of the project

We understood that the best way to achieve a commom language would be using a chatbot in Microsoft Teams and also in Microsoft Sharepoint (our intranet) and that the chatbot language should as human as possible, considering all the cultural aspects of our internal communication. The other key feature was to understand that a curator team is a main advantage, because each month new content (that Olivia does not know yet) is beign asked for and the curator team can see what are the most desired topics and this is made available in the next week or month. The other feature that is rather unique is that Olivia can begin a chat with the users, to ask for a survey to be fullfilled, if there is some pending tasks regarding that.
Used software
Inbot chatbot platform.
Difficulty of implementation
The main difficulties were related to the construction of the knowledge base, what should be a priority in each month, and with that, the lesson learned is that a curator team is fundamental. The other issue regards the API integration with other systems and applications, mainly regarding how to build a secure API integration, as some of the applications contains PII.
Project Description
In late 2021, we begun to build a chatbot for Teams to help the users to open tickets in our ITSM. But we soon saw that the users wanted more and asked more, so we partned with Inbot, a brazilian company specialized in corporate chatbots, to build Olivia, our newest and savviest employee, that knows eveything about the corporate culture, policies, commom questions and doubts. We begun with some static content regarding the most asked topics for all the areas within the company. Later on, we move to integrate using APIs with other applications in the company as the ITSM, the HR and payroll, and an internal innovation awards platform. The idea is to give the CEO or the CFO the possibility to ask Olivia for the current EBITDA or revenue of some project or region at any time using their mobile Teams app. We have a curator team that monitors all the interactions where Olivia does not "know" the answer and every month we decide which are the topics to be learned. We also developed a strong communication and change management campaign to always inform the users about the new skills tha Olivia has learned and how many interactions we are getting with her. In the roadmap, we are developing a new LLM model (similar to chatGPT) to give even better answers and also to speed up the learning process. We currently support just brazilian portuguese but we also plan to support spanish and english as well.
Project geography
The current reach is Brazil and Angola, portuguese speaking countries, but we are rolling out the spanish version next year, what will expand the reach to Peru, Panama, Dominican Republic, Argentina, Colombia, Mexico and Guatemala.
Additional presentations:
The winners of the Project of the Year 2023 competition are IT leaders from India, Malaysia, Turkey, Kazakhstan, Uzbekistan, Kyrgyzstan, Armenia, Iraq and Brazil. 

The beginning of the year was marked by a flurry of activity by various legislators to deal with artificial intelligence. The New York Times filed a lawsuit against OpenAI and Microsoft, and the UAE created an AI council to accelerate the adoption of AI systems in critical sectors of the economy.

What should a business solution be, so that it is chosen by young people, but at the same time not rejected by older professionals? Maybe beauty will save the peace in the team, because everyone wants to work in an application that looks beautiful? It is true, but User Interface (UI) is good when it is based on User Experience (UX). If harmony between UI and UX is achieved, the product will be both beautiful and user-friendly.

Choosing between cross-platform development and native development for iOS and Android is not just a technical decision, it is a strategy that can determine the future of your mobile product.

Andrey Lukin, senior PHP-developer at an international dating startup, tells us how to create and implement an internal framework into the team's workflow so that the results live up to expectations.

We use cookies for analytical purposes and to deliver you the best experience with our website. Continuing to the site, you agree to the Cookie Policy.