SUPERSTORE 007

#Past the dashboard

SUPERSTORE 007 is a Tableau Next Hackathon 2026 build that takes the most-used demo dataset in analytics and pushes it past the dashboard. Orders, returns, service cases, and social mentions all land in Salesforce Data Cloud. Account managers see the trend, ask why, create the case, and get team confirmation. Without ever leaving Slack.

The pitch is in the name. Classic Superstore tells you what happened. Bar charts, line charts, the question "why is Central down again?" SUPERSTORE 007 is licensed to act on it. The dashboard raises the question, the agent and the flow take it the rest of the way.

What it demonstrates: agentic analytics on a dataset every Tableau person already knows by heart. Same data, different outcome. The "now what" gets answered.

#How it works

Five layers, each doing one job:

• Salesforce Data Cloud. The unification layer. Orders, returns, service cases, and social mentions in one place, so the agent has context to work with.
• The semantic model. Customer Health Intelligence: nine calculated fields, seven metrics, and a business preference that says what "unprofitable" means. Profit below zero. Field descriptions in business language, so the agent answers in it.
• Tableau Next. The surface where trends become questions. Same Superstore visuals you've seen a thousand times, wired to a Data Cloud source instead of a flat extract.
• Agentforce. The out-of-box analytics agent, plus one custom action: create the case.
• Flows. A High-priority Account Review case linked to the account, then an alert in #customer-alerts.

Slack is where it all lives. No tab-switching, no "did anyone follow up on that?"

The move here: stop building dashboards that end in questions. Build for the action on the other side. Any operational dataset works, as long as "what should we do about it" has an answer.

#Try this

If you've got a Salesforce dev org and some time, sample the beats. Five of them:

1. Pick the boring dataset on purpose. Superstore. Or whatever your org's equivalent is, the one everyone already understands. Nobody reads Superstore for novelty. The action layer is the whole story.
2. Land it in Data Cloud, not in a flat extract. Even one source is enough to start. The unification matters more once a second source shows up: service cases, support tickets, anything that gives the agent context the dashboard alone can't carry.
3. Wire one Tableau Next view to it. Doesn't need to be pretty. Needs to be where the question gets asked.
4. Build one Agentforce action that picks up that question. Start narrow: one trend, one follow-up, one case template. Resist the urge to make the agent omniscient on day one.
5. Hand the case to a Flow that creates the record and posts to Slack. This is the step a team keeps.

The whole loop is the bit. You can extend any layer afterwards (richer sentiment, smarter routing, multi-step agent reasoning), but the loop has to close first. Question on the dashboard, action in Slack, no tabs in between.

#What it taught me

The dashboard doesn't have to be impressive. The handoff does.