SPACE

Appearance

Using the assistant ​

The SPACE AI Assistant provides a conversational interface for working with SPACE, IBM Planning Analytics models and data, and other supported connections. Start with a question, investigation, or task, then continue with follow-up questions as the assistant gathers context and returns results.

You do not need to choose a SPACE module or tool before starting. Describe what you want to understand or accomplish, and the assistant can use the SPACE-controlled tools available for the current conversation where appropriate.

Connection context and AI profiles ​

The assistant interface provides connection context and an AI profile selector.

The selected connections determine which configured environments and data sources are available to the conversation. Depending on the task, SPACE AI can work with IBM Planning Analytics connections and other supported connection types, such as Google BigQuery.

The selected AI profile determines the configured AI provider and model used for the conversation. Organizations are not tied to a single AI vendor or model: administrators can configure supported providers and models and make appropriate profiles available to users. For example, an environment may provide profiles using OpenAI, Anthropic, or different models for different use cases.

Changing the AI profile changes the configured provider or model used for reasoning without changing how you interact with SPACE AI or the SPACE tools and connections available to the conversation.

Administrators create and manage available profiles under AI configuration.

Supported IBM Planning Analytics and Google BigQuery connections can use SSO, allowing SPACE AI to access data using the current user's individual authentication and permissions. Users do not need to provide their source-system credentials to the AI Assistant or to SPACE.

Access to connections and SPACE capabilities remains governed by SPACE permissions, while the permissions of the connected system continue to determine what data and objects the user can access. See Connections and Access for configuration details.

Asking questions ​

Start by describing the result you want rather than trying to identify the SPACE tool that should be used.

For example:

  • Explain how revenue is calculated in this model.
  • Show me the processes that update this cube.
  • Analyze this data and explain the largest variances.
  • Why did this TurboIntegrator process fail?
  • How do I access a configured TM1 connection from Python?

Include important context when it is relevant, such as the cube, process, environment, time period, or error you are investigating. If SPACE AI needs additional information, it can ask for clarification.

See Examples for additional use cases and sample prompts.

Follow-up questions ​

SPACE AI maintains conversation context, so follow-up questions can build on information already discussed or retrieved.

For example, after asking the assistant to explain a rule, you can continue with questions such as:

  • Which cubes does it reference?
  • Explain the feeders in more detail.
  • Are there any potential performance issues?

You generally do not need to repeat the complete original question. The assistant can use the current conversation to resolve references and continue the investigation.

When a follow-up changes to a different environment, connection, object, or subject, state that change clearly so the assistant does not rely on an incorrect assumption from earlier in the conversation.

Longer conversations ​

For longer conversations, SPACE may summarize older messages into compact conversation memory. This preserves important context while keeping the amount of conversation history sent to the AI model manageable.

Original messages remain stored in SPACE. Conversation compaction changes the context supplied to the AI model for subsequent responses; it does not replace the stored conversation history.

If an important detail from much earlier in a long conversation becomes relevant again, restating it can help ensure that it is considered.

Reviewing results ​

Review the assistant's assumptions, source information, and results before relying on them for important business decisions or production changes.

When SPACE AI uses available tools or documentation, continue the conversation if you need more detail, want to verify an assumption, or want the assistant to investigate a related question.

For development or operational changes, use the normal SPACE and IBM Planning Analytics review and testing processes appropriate to your environment.

Formatted responses ​

SPACE AI responses support Markdown formatting, including:

  • Headings
  • Lists
  • Tables
  • Inline code
  • Code blocks
  • Links

This allows model explanations, rule and process reviews, data analysis, troubleshooting results, and documentation guidance to be presented in a structured, readable format.