Contextual AI prompts inside the analytics control plane

Product Update · JoomAI Release

Raw analytics dashboards require merchants to perform two distinct cognitive tasks: first, evaluate numerical variance across metrics; second, synthesize those numbers into a commercial decision.

To bridge this gap directly at the point of analysis, we have integrated contextual AI actions across our analytics dashboards. Merchants can now launch structured prompts into JoomAI, Claude, or ChatGPT in a single click without leaving their operational view.

Reducing friction at the metric layer

Traditional analytics platforms report state but leave interpretation entirely to the user. When evaluating complex data—such as category seasonality, revenue dynamics, or seller monopolization rates—merchants often struggle to translate raw charts into specific inventory or pricing actions.

By placing targeted "Explain" and "Explain seasonality" actions inline next to core data widgets, we eliminate the friction of manual prompt construction. The interaction flows natively through four stages:

• Raw Analytics Metric: The merchant views localized performance or category data on screen.
• One-Click Action: Selecting an inline trigger activates the prompt generator.
• Pre-Structured Prompt: The system automatically binds relevant telemetry to a targeted query context.
• Reasoned Output: The selected model returns actionable analysis directly to the user.

For instance, evaluating category opportunity automatically constructs a detailed prompt breaking down competitive intensity, seller medal share, average revenue per merchant, and growth trajectories:

"Explain the opportunity level for the cookware category. Detail each underlying factor: monopolization (15.3%), sellers with medals (18.6%), average revenue per seller (R$ 7.6k), and revenue dynamics (16%)."

Interoperability across model environments

Merchants have different operational workflows and model preferences. Rather than restricting analysis to a single chat interface, the action menu offers direct routing to multiple execution environments:

• JoomAI: Native execution optimized for internal platform workflows.
• Claude / ChatGPT: External routing leveraging the underlying data connector to pipe live marketplace data directly into third-party AI clients.

This architecture treats the analytics dashboard as a launchpad, allowing sellers to inspect metrics inside whichever model environment they prefer for daily work.

Instrumenting user intent for product strategy

Beyond immediate utility for merchants, inline prompt triggers function as an telemetry signal for our product engineering team.

Every action invocation provides structured data on:

• Metric pain points: Identifying which specific data widgets (e.g., seasonality vs. seller density) require the most frequent interpretation assistance.
• Model distribution: Tracking real-time user preference between native execution (JoomAI) and external clients (Claude, ChatGPT).

These interaction signals directly inform our core product roadmap, guiding future prompt templates, data connector capabilities, and interface refinements.

By embedding reasoning tools directly into the analytics interface, we shift dashboards from passive reporting tools to active decision environments—increasing platform engagement and deepening how merchants interact with their store data.