AI Agent Overview

What is the Stackless Agent?

The Stackless Agent is a conversational workspace for exploring data and building analytics assets. You can ask plain-English questions, but the Agent is designed to do more than return a temporary answer. It can create saved artifacts, draft models, build dashboards, inspect lineage, and continue work across conversations.

[picture 1. Agent workspace with conversation history, chat, and artifact panel]

What the Agent can do

The Agent can help with:

• Discovering available sources, tables, models, metrics, and dashboards.
• Running read-only warehouse analysis within your role's access.
• Explaining metric definitions and dashboard lineage.
• Creating table artifacts, chart cards, finding cards, and briefs.
• Building private dashboards and helping publish them when ready.
• Drafting or updating Transformation Models.
• Creating or previewing Semantic Models.
• Checking downstream impact before changing a model.
• Working with uploaded CSV or XLSX files.

The Agent's context

The Agent has access to the Stackless Catalog, your visible dashboards, your visible models, and the data schemas available to your role. It also knows conversation context, selected artifacts, uploaded files, and active dashboard drafts.

It does not have unlimited access. If your role cannot see a schema or dashboard, the Agent should not use it.

The artifact workspace

The right side of the Agent page is the artifact workspace. It shows durable outputs created or attached in the current conversation:

• Briefs.
• Finding cards.
• Table artifacts.
• Chart cards.
• Private dashboard drafts.

Artifacts are how useful Agent work survives beyond a chat response. A table artifact can be exported. A chart card can be copied to a dashboard. A dashboard can be published for recurring reporting.

[picture 2. Artifact workspace showing a brief, table, chart card, and draft dashboard]

How a typical Agent workflow works

1. You ask a question or request a build.
2. The Agent searches the Catalog or existing models.
3. The Agent runs safe analysis or drafts the needed object.
4. The Agent returns an answer, artifact, model draft, or dashboard draft.
5. You review the result.
6. If the output should persist, you save, attach, publish, or promote it.

Example workflows

Explore existing reporting

Ask:

• "What dashboards do I have access to?"
• "Explain how this dashboard's revenue metric is calculated."
• "Which upstream models feed this KPI?"

The Agent should inspect existing dashboards, metric definitions, and lineage before answering.

Create a table artifact

Ask:

• "Show a table of revenue by channel for the last 30 days."
• "Save the result as a table artifact."

The Agent can create a table snapshot with columns, rows, source query metadata, row counts, and masking where needed.

Build a dashboard

Ask:

• "Build a dashboard for weekly revenue, orders, AOV, and active customers."
• "Add channel and country filters."
• "Create a table tab for campaign performance."

The Agent should use existing Semantic Models when available. If the right model does not exist, it may propose Transformation Model or Semantic Model work first.

Create model-backed reporting

Ask:

• "Create a Transformation Model for customer-level daily revenue."
• "Preview it and show sample rows."
• "Create a Semantic Model on top of it."
• "Build a dashboard from the new model."

This moves from raw data to governed reporting instead of producing one-off SQL.

What the Agent should not be used for

The Agent is not a substitute for business review. Review outputs before publishing, especially when:

• A model changes a published reporting dependency.
• A metric is used in executive reporting.
• A dashboard is shared with a broader role.
• The analysis uses uploaded files.
• The result includes sensitive data.

The Agent can make mistakes. Verify important results with the underlying table, model, dashboard, or Stackless team.

Next steps

• Using the Agent explains the conversation workflow.
• Artifacts explains saved outputs.
• Building Dashboards explains dashboard creation.
• Best Practices gives prompt patterns that produce better results.