Having the right structure and tools in place to build virtual assistants is often as important as actually doing it. Kore.ai provides everything you need to create, train, test, and host virtual assistants using one consistent process – even across multiple teams, locations, and release dates.
Defining Virtual Assistant Lifecycle Management
Virtual assistant lifecycle management is a systematic framework that defines all the steps involved in planning, building, testing, and deploying virtual assistants. Every step has its own clearly defined process and deliverables that feed into the next step in the framework.
Virtual assistant lifecycle management provides stakeholders with a common vocabulary and a shared set of goals and responsibilities. This approach helps simplify project planning, staffing, and management, and reduces the risk of bugs and rework as each phase is validated by stakeholders and other trusted team members.
Defining Virtual Assistant Lifecycle Management
Virtual assistant lifecycle management is a systematic framework that defines all the steps involved in planning, building, testing, and deploying virtual assistants. Every step has its own clearly defined process and deliverables that feed into the next step in the framework.
Virtual assistant lifecycle management provides stakeholders with a common vocabulary and a shared set of goals and responsibilities. This approach helps simplify project planning, staffing, and management, and reduces the risk of bugs and rework as each phase is validated by stakeholders and other trusted team members.
Phases of Virtual Assistant Lifecycle
Phase 1: Define Blueprint
Blueprinting involves determining the overall scope of the project, collecting all functional and non-functional requirements, and defining features against the requirements identified. This phase can be done in-house or in conjunction with a team of project implementation experts. Design and development can begin once key stakeholders sign off on these plans.
Phase 2: Design Conversation
This phase involves designing initial conversational experiences. Typically, this step would require storyboarding or lengthy conversational scripts. With Kore.ai, however, teams can work together to create full user-bot interactions by typing in messages within a conversational designer, as well as commenting on, adding notes, and assigning tasks related to interactions – enabling collaboration and information dissemination between designers and stakeholders.
Phase 2: Design Conversation
This phase involves designing initial conversational experiences. Typically, this step would require storyboarding or lengthy conversational scripts. With Kore.ai, however, teams can work together to create full user-bot interactions by typing in messages within a conversational designer, as well as commenting on, adding notes, and assigning tasks related to interactions – enabling collaboration and information dissemination between designers and stakeholders.
Phase 3: Build Solution
Once initial conversations have been designed, individual components can be developed. Front-end components include the conversational interface itself – translating user inputs into specific actions you want the virtual assistant to take and vice versa. Back-end components handle volume, performance, and integration to systems and web services. Kore.ai provides a comprehensive task execution framework that makes it easy to match requirements from your blueprints to virtual assistant functions.
Phase 4: Implement
This phase refers to implementing the initial build and conversational flows, as well as the initial language model. This is typically done on an iterative basis, with each build being used to create demos for key stakeholders. While such development traditionally requires extensive coding, Kore.ai provides a dialog builder for creating complex, human-like experiences without writing code.
Phase 4: Implement
This phase refers to implementing the initial build and conversational flows, as well as the initial language model. This is typically done on an iterative basis, with each build being used to create demos for key stakeholders. While such development traditionally requires extensive coding, Kore.ai provides a dialog builder for creating complex, human-like experiences without writing code.
Phase 5: Train and Test
Next, training is performed so the virtual assistant can interpret diverse human expressions. To simplify this onerous process, Kore.ai provides three different training engines, each with its own methodologies and advantages. The virtual assistant is then deployed to a test environment for user acceptance testing and sign-off. Here, Kore.ai provides batch testing capabilities and other tools to gauge the effectiveness of the initial training and determine the model’s overall accuracy, as well as validate it will continue to work as expected as changes are made.
Phase 6: Go-Live
At this point, the virtual assistant is ready for deployment to a production environment, and ongoing governance is needed. Kore.ai provides extensive bot and user management, including custom roles and permissions, selective publishing, maker-checker controls, and more.
Phase 6: Go-Live
At this point, the virtual assistant is ready for deployment to a production environment, and ongoing governance is needed. Kore.ai provides extensive bot and user management, including custom roles and permissions, selective publishing, maker-checker controls, and more.
Phase 7: Post Go-Live Support
Virtual assistant development is an iterative process. To facilitate development, Kore.ai provides a comprehensive analyze and learn module for reviewing, refining, and retraining NLU models, as well as for performing root cause analysis on failure scenarios. Coupled with extensive testing tools, full chat transcripts, and detailed analytics, Kore.ai enables teams to perform hypercare, measure bot stability, and enhance their virtual assistants over time.
Virtual Assistant Lifecycle Management Support
Versioning
Versioning allows you to create multiple versions of the virtual assistant at various stages of the life cycle. Changes to the original task do not go into effect until they have been finalized and approved, so adjustments can be made without impacting ... current operations. The versions of the virtual assistants can be automatically created at the time of publishing. You are also allowed to create different versions during development. Our platform stores different versions of the virtual assistants, which can be retrieved and restored as and when required. You can roll back to the previous versions and never lose any changes made.Cloning is also provided so a new separate task can be built upon an existing task without being tied to the original version.
Provide Role-Based Access
Not all virtual assistants should be shared with everyone. Kore.ai provides administrators tools to manually assign virtual assistants to specific enterprise users and testers. When combined with role-based access controls, this ensures only the right personnel ... have access to potentially sensitive virtual assistant functions and data. This also allows virtual assistants to be deployed incrementally by gradually increasing the user base
Test on Multiple Environments
Each phase in the lifecycle should take place in separate environments, so developers and testers can work in tandem. Kore.ai provides various dev, test, and production environments, as well as a sandboxed runtime environment to test virtual ... assistant configurations. Regardless of the environment, the same virtual assistant can be used for both development and production, eliminating the need to rebuild a new version of it with every iteration and allowing for further development without impacting end users.
Share via Import and Export
In addition to built-in versioning, Kore.ai supports versioning in external repositories through importing and exporting virtual assistants. Also individual components, such as language configurations can be imported or exported. Relevant files ...can be imported, either incrementally or in full, into the same virtual assistant for the purposes of upgrading it or can be used to replicate the bot in another environment. In addition, Kore.ai provides release management APIs to automate virtual assistant deployments across physical environments.
Benefits of Virtual Assistant Lifecycle Management
