Software Engineer - ACB Chatbot

*Please note: the service contract for this position will not be concluded with Henkel AG & Co. KGaA but with an external party”.

Projektname /project name

ACB Chatbot

Projektbeschreibung /project description

The service is requested as part of the project ACB Chatbot (ACB = Construction Adhesives Unit of Henkel). The project has the purpose of building a fully functional virtual assistant (chatbot) for the Construction Adhesives customers (craftsmen) and will be initially deployed on the websites and applications (where applicable), as well as Facebook Messenger, but will later be included in other messaging applications.

Leistungsbeschreibung / task description

The service of the contractor is delivered using an agile working method. External resources are needed as there is no internal staff with the required expertise in the following areas:

• Development of an ACB custom chatbot solution that will run on the Cognigy platform (Cognigy is an external vendor for chatbots; Henkel’s Laundry and Home Care Unit already owns a license and this chatbot will run on that platform as well)

Therefore, the external contractor is in a unique position and performs significantly different tasks than the internal employees.

One sprint consists of five weeks and there is a weekly jour fixe. During these meetings, the team discusses the current requirements and the contractor independently performs the following tasks:

Chatbot Dialog Design within the Cognigy.AI platform

• Create the chatbot in the Henkel Cognigy.AI platform
• Creation of a user-friendly, natural conversation flow
• Design information architecture for the chatbot
• Training of the chatbot regarding intents and synonyms 
• Implementation of the bot
• Configure multi-language localizations of the bot which are specified in the project plan
• Pilot in one country (probably Russia) in 2021 and scale up to as many more countries as possible until the end of 2021 (the plan is up to 3 more); then scale up globally throughout 2022
• Conception of additional channels; possible channels include Facebook Messenger, WhatsApp Business, Viber, and Telegram (the MVP will include Facebook Messenger, then the plan is to include the rest as the chatbot evolves)

Chatbot Data Management Design

• Define data structures for the chatbot
• Define technical structures for the bot to use
• Create human-manageable data structures for data exchange and providing by Henkel Employees 
• Independently create Excel files that will allow Henkel brand employees to write down data like help texts or providing Service Center Hotline phone numbers;
• Transform these data files for input for the chatbot implementation; the specific formats and processes are determined as part of the project plan.
• Define and implement data usage in the bot
• Conception of connecting to other Henkel data sources like PIM, mail services etc. (these integrations will be implemented at later stages and will not be part of the MVP stage)
• The pilot implementation will lead to knowledge about what kind of data sources should be included. Only then will we be able to find Henkel systems that provide this data, and then we will know if we can access that data and how. Listing the systems is not possible at this point in time.

Platform Management and Training

• Set up all necessary artifacts in the Cognigy.AI platform
• Define bot integration for all channels (the MVP will include Facebook Messenger, then the plan is to include the rest as the chatbot evolves)
• Train both the brand team and Henkel dx Smartbot Factory Team to use and implement the Cognigy.AI platform
• The training will include a short session for the brand team (up to 5 participants) to add intents and lexicon entries for enhancing the bot – 2 hours max. 
• DX training will have a starter session of less than 4 hours to explain the platform and create simple demo bots; beyond that, it is planned to be part of the implementation.
• Create a PoC reporting in PowerBI based on the platform’s Data analytics interface.
• The details of the final report need to be discussed during the project and after launching the bot. The initial proof of concept reporting will show information regarding bot usage (number of users and sessions) and the top questions, as well as listing how many user inputs were not answered by the bot.

Technical Project Management

• Communications with external localization/translation agency for international rollout
• Communications with external Cognigy.AI technical support and architect to add platform features necessary for Henkel

Chatbot Dialog Implementation within the Cognigy.AI platform

• Training of the chatbot regarding intents and synonyms 
• Implementation of a user-friendly, natural conversation flow according to Henkel best case scenarios provided to the contractor in advance
• Implementation for multi-language localizations of the bot
• Pilot in one country (probably Russia) in 2021 and scale up to as many more countries as possible until the end of 2021 (the plan is up to 3 more); then scale up globally throughout 2022
• Additional channels discussed include Facebook Messenger, WhatsApp Business, Viber, and Telegram (the MVP will include Facebook Messenger, then the plan is to include the rest as the chatbot evolves)
• Independently create and manage test and integration environments

Chatbot Data Management

• Implement data storage and services for the chatbot
• Transformation of Henkel data formats to chatbot data sources
• Define and implement data usage in the bot
• Platform Management and Training
• Manage Cognigy.AI platform in terms of users and projects: 
• User management
• Cognigy Agent administration
• Create technical users and API keys for programmatic access and reporting 
• Create and update chatbot endpoints for inclusion on web pages and Facebook Messenger
• Additional Project Tasks
• Communications with external Cognigy.AI technical support and architect to add platform features necessary for Henkel

The service provision of the contractor has the goal of creating a consumer helping chatbot that fulfils the given use cases and enabling Henkel brand team and Henkel dx to understand the platform, maintain parts of the chatbot themselves and be knowledgeable about what additional features should be added to the chatbot. 

Timelines

The following timelines are to be adhered to by the contractor during the provision of the service:

The contractor estimates approximately 25 days of work for implementing the MVP version of the bot. Implementation of subsequent features are specified in the project plan.