Kore.ai introduces SDLC best practices to the world of virtual assistants so you can design, create, train, test, and host natural and engaging conversational experiences. Our platform empowers you to effectively and efficiently build and ship virtual assistants anywhere using one consistent process for all your teams.
Why effective virtual assistant development requires iteration
Creating and maintaining effective virtual assistant is an interative process. There are almost always going to be new tasks, dialogs, and bug fixes waiting to be designed, developed, and deployed. The best assitants are those that are constantly evolving, that is that they come about when you build a bot, get it out into the wild to see how people use it, and then use that data to test, iterate, test, and iterate some more.
Why businesses need SDLC support for virtual assistants
SDLC support provides all stakeholders with a common vocabulary and a standard “definition of done” as well as defined processes, communication flows, and responsibilities for IVA creators, analysts, testers, and admins – ensuring that everything is well documented and that everyone knows exactly what they need to do during each step of the bot building process.
Versioning and release management provides a single source of truth for releasing complex enterprise virtual assistants at scale, allows you to easily pinpoint the root cause of potential failures, and allows you to quickly update, revert, or rollback problematic releases.
Kore.ai provides full SDLC support, including support for development lifecycles across multiple teams with multiple release dates and support for companies that have different physical environments for each SDLC stage. Our platform also provides the ability to manage development lifecycles across various dev, test, and production environments and a sandboxed runtime environment to test virtual assistant configurations.
Task-Level Version and Release Management
Publishing Virtual Assistant Tasks
Kore.ai provides a seven-step process for publishing new or updated tasks, including clearly defined inputs and outputs from one step to the next, that helps clarify roles and responsibilities among IVA creators, analysts, and project managers. This process, in conjunction with version management, allows VA creators to use the same assistant for both development and production – eliminating the need to copy, build, and deploy a new version of the assistant definition with every iteration and allowing you to develop new tasks without impacting the end users.
Upgrading and Versioning
Visually track and monitor major and minor changes to your virtual assitant and task configurations with version numbers that are automatically and incrementally generated. Kore.ai allows you to define major and minor releases and determine if task upgrades are mandatory or optional for bot end-users.
Kore.ai ensures proper permissions are maintained at all times between the development, testing, and deployment phases of the SDLC by requiring that bot creators resubmit their tasks for approval whenever major changes are made to the VA configuration.
Importing and Exporting
Kore.ai supports external versioning in your prefered repository by allowing you to export and import complete virtual assistant or individual assistant components, such as dialog tasks or ML configurations. IVA definition files can be imported, either incrementally or in full, into the same assistant for upgrading purposes or can be used to replicate the bot to another environment.
Kore.ai also provides release management APIs to automate IVA deployments across physical environments.
Any and all changes made to a VA definition, including tasks, is recorded in an audit trail log and assistant change log. This provides all stakeholders with a chronological list of modifications to the virtual assistant and a single source of truth, and includes comprehensive details that answer the who, what, when, and how for each change.