Skip to main content

Deploying and Configuring the CI Engine

The Collaborative Intelligence Engine can be deployed as a Data Experience Definition package utilizing Cinchy's DXD tool. The package contains all of the tables, queries, syncs, and other objects needed to run the CI Engine. Using DXD, these objects will be exported from the CI source environment and imported into your target environment.

Prerequisites​

Prior to deploying the Collaborative Intelligence Engine, take note of the following requirements:

  • The Cinchy Platform v5.0+ (SQL Server only)
  • The CI Engine Data Experience Definition package
  • Cinchy DXD must be installed in the source environment

Data Experience Definition Package​

Upon deploying the CI Engine's Data Experience Definition package, the following objects will appear in your environment. You can cross-reference this list after deployment to confirm that the CI Engine package installed correctly.

  • Models:
    • Collaborative Intelligence Engine V1.0.0
  • Domains:
    • The [Collaborative Intelligence] domain will be created, if not already present in your environment.
  • Tables:
    • [Collaborative Intelligence].[CI Recipes]
    • [Collaborative Intelligence].[CI Insights]
    • [Collaborative Intelligence].[AI Models]
    • [Collaborative Intelligence].[AI Model Execution Log]
    • [Collaborative Intelligence].[Formats]
    • [Collaborative Intelligence].[Model Entities]
    • [Collaborative Intelligence].[Model Attributes]
    • [Collaborative Intelligence].[Packages]
    • [Collaborative Intelligence].[AI Safety Filters]
    • [Collaborative Intelligence].[CI Quality Levels]
  • Saved Queries:
    • Sync CI Insights Now
    • Save Model Output JSON
    • Refresh CI Insights Every N Seconds
    • Refresh CI Insights Now
    • Log AI Model Execution
    • Calculate Model Input JSON
    • Infuse CI Insight Now
    • Infuse CI Insights Now
    • Request Refresh CI Insights Now
    • Reset Update CI Now Flag
    • Save Model Input JSON
    • Get CI Recipe Dependencies
    • Model Needs Execution
    • Reset Execute Now Flag
    • Get CI Recipe Tooltips
  • Groups:
    • CI Engine Admins: This group gives edit access to all of the objects in the package.
    • CI Engine Builders: This group gives design access to the objects, as well as edit access to specific columns in the package tables.
    • CI Engine Users: This group gives users access to create questions, as well aa edit the questions they have access to.
  • User Defined Functions:
    • askCinchy
    • logAIModelExecution
  • Data Sync Configurations:
    • [Collaborative Intelligence].[Refresh CI Insights Every N Seconds]
    • [Collaborative Intelligence].[Refresh All Now]
    • [Collaborative Intelligence].[Refresh Now]
    • [Collaborative Intelligence].[Sync CI Insights Now]
    • [Collaborative Intelligence].[Infuse Now]
    • [Collaborative Intelligence].[Infuse All Now]
    • [Collaborative Intelligence].[Refresh CDC - Comments]
    • [Collaborative Intelligence].[Execute Now]
    • [Collaborative Intelligence].[Sync CI Recipe Dependencies]
    • [Collaborative Intelligence].[Sync Recipe Tooltips]
  • Reference Data
    • [Collaborative Intelligence].[AI Models] table
    • [Collaborative Intelligence].[Answer Formats] table

How to Deploy​

  1. Using Cinchy DXD, export the Collaborative Intelligence Engine data experience package out of the Cinchy source environment.
  2. Using Cinchy DXD, install the CI Engine data experience package into your target environment.
  3. Validate that the deployment was successful by cross-referencing that the above list of tables/queries/syncs/etc. is available in your target environment.
  4. Navigate to the [Collaborative Intelligence].[Formats] table and approve all of the pending data.
  5. Navigate to the [Collaborative Intelligence].[AI Models] table and approve all of the pending data.
  6. The [Collaborative Intelligence].[CI Recipes] table relies on Bearer Token/Personal Access Token authorization to produce results that adhere to Cinchy's federated access controls. (I.E., answers will only return insights based on data that the token has access to.) Navigate to the [Cinchy].[Secrets] table and input whichever tokens that you want available when configuring questions for the CI Engine.
  7. Navigate to the Connections Experience. For each of the data syncs deployed via the DXD, ensure that the Job tab is set to Enabled.

Configuring the CI Engine

There are various ways that you can configure your CI Engine to optimize the insights being returned to you. These configuration include:

  • Specifying the AI model
  • Defining different answer structures
  • Configuring the frequency in which repeatable questions are posed to the CI Engine
  • Swapping authorization/bearer tokens

Specifying the AI Model​

The Model column of the [CI Recipes] table is used to specify which AI Model will be used to answer the given prompt. This is configured on a recipe-by-recipe basis.

To specify the AI Model, select a value from the Model column drop down.

::: tip Additional models can be added to the engine by a CI Engine Admin via the [Collaborative Intelligence].[AI Models] table. :::

Answer Structures​

Question Frequency​

Authorization Tokens​