Concepts → AI Containers

About AI Containers

An AI Container is a centralized workspace that groups approved data assets, instructions, and Knowledge Store resources so that Incorta Nexus can answer questions with consistent business context. AI Containers provide a reusable, governed context for Nexus conversations, eliminating the need to rebuild it each time.

Within Incorta, an AI Container acts as a context unit. Instead of asking Nexus to reference many separate assets scattered across your workspace, you bundle related Business Views, Dashboard Insights, instructions, and Knowledge Stores into a single container. When a user attaches an AI Container in a Nexus chat, Nexus uses all the container's assets and instructions to inform its responses, providing a consistent and predictable conversation context.

For information on configuring AI Containers, refer to References → AI Containers Configuration Guide.

Key characteristics of AI Containers

• Governed: Follow your existing Incorta sharing model. Owner, creator, or administrator can grant access through role-based permissions.
• Reusable: Create once, attach in multiple Nexus conversations. No need to reconfigure context for each chat.
• Flexible: Rename, delete, export, and import containers to manage your context library.

AI Containers in Nexus

AI Containers are surfaced to users as a new context type in Incorta Nexus. In a Nexus chat, users can select the Context menu, search for an AI Container by name, and attach it to the conversation. Once attached, Nexus uses the container's data assets, instructions, and Knowledge Store resources to answer questions within that chat session.

The relationship between AI Containers and Nexus is one-directional: AI Containers exist to provide context to Nexus conversations. Nexus does not modify or influence the container's configuration.

Note: When the Include in Nexus toggle is disabled, AI Containers will be hidden from Nexus and will not appear in the context picker, even if the user has access to the container. The container remains available in Content for configuration and sharing purposes.

Data assets

Only Nexus-enabled assets are eligible for inclusion in an AI Container, including:

• Business Views: The primary unit of data context. Multiple Business Views can be added to a single container.
• Dashboard Insights: Insights from dashboards, respecting any Default Applied Dashboard Filters. Multiple insights can be added from the same dashboard.
• Knowledge Stores: When enabled, Knowledge Stores can be added as instruction context or as standalone resources. This allows containers to reference Knowledge Store entries for domain-specific guidance and data reference.

Instruction use cases

AI Container instructions define how Incorta Nexus interprets business context, applies organizational standards, and responds to user questions. Common use cases include:

• Metric and calculation definitions: Define how a business metric should be calculated when it is not a straight column reference.
  • For example, instruct Nexus to calculate churn rate as customers lost divided by customers at the start of the period or define an active customer as one who has placed an order within the last 90 days.
• Disambiguation rules: Resolve ambiguous business terms by defining the default interpretation when multiple fields could answer the same question.
  • For example, if a user asks for revenue without qualification, use Net Revenue instead of Gross Revenue, or interpret sales as Bookings rather than Shipped Revenue unless the user specifies otherwise.
• Default scope and filters: Apply business assumptions when users do not provide sufficient context.
  • For example, default to the current fiscal quarter if no date range is specified, or automatically exclude test accounts and internal orders from results.
• Comparison and benchmarking: Standardize how performance is evaluated.
  • For example, compare regional performance with the same period in the previous year rather than the previous period, or identify metrics that fall more than 15% below target as underperforming.
• Terminology and glossary: Define domain-specific terminology to ensure consistent interpretation.
  • For example, define lead time as the number of days between order placement and delivery or safety stock as the inventory buffer maintained to prevent stockouts.
• Units, formatting, and rounding: Standardize how values are presented.
  • For example, always display currency in USD rounded to the nearest thousand or display percentages with one decimal place.
• Guardrails and scope limitations: Define topics that Nexus should not answer or redirect on.
  • For example, do not provide individual employee compensation data and indicate when a question falls outside the AI Container's available data rather than making assumptions.
• Clarification behavior: Specify when to ask versus when to assume.
  • For example, if a user asks for the top customers without specifying a metric, ask whether they mean by revenue or order count before answering.

Business use cases

AI Containers provide reusable, governed context for Incorta Nexus conversations. Common use cases include:

• Sales analytics: Combine sales Business Views, dashboard insights, and business instructions to provide consistent context for analyzing sales performance, regional trends, and forecasts.
• Financial reporting: Group financial data with accounting policies and calculation guidelines stored in Knowledge Stores to support budget analysis, expense reporting, and financial forecasting.
• Marketing campaign analysis: Combine campaign data with marketing glossaries and business instructions to analyze campaign performance, customer engagement, and return on investment (ROI).
• Product analysis: Group product, inventory, sales, and customer data with business rules and naming guidelines to ensure consistent analysis across product-related conversations.