Workflows for Data Analysts

This guide walks through the workflow a data analyst follows inside Treasure AI Studio — from an open-ended business question to a polished, exportable set of deliverables. Everything happens in a single conversation: you steer the analysis, and the AI handles the tooling.

Enable data analysts to go from "What happened to our conversion rate last quarter?" to a validated, stakeholder-ready set of charts and exports in a single session — without switching between a query editor, a visualization tool, and a slide builder.

You arrive Monday morning to a Slack message from the VP of Product: "Conversion rate seems down. Can you pull together a Q1 performance overview for the leadership meeting at 2 PM?"

Normally this means: open the query editor, remember which tables hold conversion events, write and debug SQL, copy results into a spreadsheet, build charts, and paste them into slides.

In Studio, it's one conversation. You describe the goal, the AI explores the data, writes and runs queries, renders charts, and generates the deck. You steer the analysis; the AI handles the tooling.

Before you start, set up your environment so the AI has the right context and capabilities:

Tell the AI what you're looking for, not which tables to query:

I need to analyze Q1 conversion performance. What data do we have
related to user signups, purchases, and page views?

The AI explores your account to find relevant data:

1. Lists databases with tdx databases
2. Lists tables in the relevant database with tdx tables
3. Inspects promising table schemas with tdx describe
4. Summarizes what's available and proposes a query approach for you to confirm

Each command appears as an expandable tool call in the chat, so you can see exactly what the AI inspected before it writes any SQL.

Confirm the approach and let the AI write the SQL:

Yes — calculate the weekly conversion rate (signups to purchases) for Q1,
broken down by traffic source.

Guided by the active skills, the AI constructs a Trino query — using td_interval for Q1 date filtering and partition pruning for performance — and runs it with tdx query. The full query is visible in the tool call, and the results render as an interactive table. If the numbers look off, ask the AI to adjust the logic and re-run.

Ask for a chart:

Show me the weekly conversion trend as a line chart, with a separate
line for each traffic source.

The AI formats the query results and generates a self-contained HTML chart, which opens in the file panel. Charts are interactive — hover for exact values. Ask follow-ups to build up several views:

The paid drop is steeper than I expected. Add a scatter plot of spend
vs. conversion for paid traffic.

The AI runs a follow-up query and renders the second chart in the file panel.

If a query runs too long, ask the AI to optimize it:

That last query took too long. Can you optimize it?

Guided by the trino-optimizer skill, the AI analyzes the query, identifies missing partition pruning or expensive joins, and rewrites it using optimization patterns — for example, CTAS for intermediate results, APPROX_DISTINCT for high-cardinality counts, or REGEXP_LIKE in place of multiple LIKE clauses. It re-runs the query and reports the speedup, with the rewritten query visible alongside the original so you can see exactly what changed.

Generate the deliverables for the meeting:

Create a PowerPoint deck with these findings — include the charts,
key takeaways, and a recommendation slide.

The AI assembles a PPTX (title, executive summary, charts, and recommendations) and writes it to its file space; a file card appears in the chat with Preview and Download buttons. Ask for more formats the same way:

Also export the raw data as an Excel file, with separate sheets for
the weekly metrics and the source breakdown.

The Q1 overview is one pattern. Here are other common analyst workflows in Studio:

Audit the customer-profiles table for data quality issues —
check for nulls, duplicates, and stale records.

The AI inspects the schema with tdx describe, runs queries to count nulls per column, detect duplicate IDs, and find stale records, then generates a bar chart of null rates and a prioritized summary of remediation steps.

The daily ETL workflow failed last night. What happened?

The AI lists recent runs with tdx wf sessions, reads the failed attempt's logs with tdx wf logs, identifies the root cause, and proposes a fix to the workflow definition.

I just got access to a new clickstream table. What's in it, and is it
useful for attribution modeling?

The AI inspects the schema, profiles row counts, date ranges, and key cardinalities, visualizes the distribution of event types, and gives you an assessment with a recommended join strategy.

Throughout this workflow, the AI makes decisions you can see and influence:

The AI follows active skills (trino, time-filtering, trino-optimizer) for platform-specific patterns, but you always have the final say. Every tool call is expandable — click to see the exact command and its result.

At the end of this workflow, you have:

• A validated analysis — every number traces back to a specific query visible in the chat
• Interactive charts — with tooltips and legends, ready to present on a screen-share
• Exportable deliverables — a PPTX deck and an XLSX data file ready to download
• A record of the session — the full conversation captures your analytical process and the exact queries run

All produced in a single session, without switching between a query editor, a BI tool, and a slide builder.

After following this workflow, you can:

• Explore an unfamiliar database using natural language and the AI's tdx describe calls
• Request and receive a Trino query with TD-specific functions, and see it run
• Generate a chart from query results and change its type
• Optimize a slow query with the trino-optimizer skill
• Export results as PPTX and XLSX for stakeholder delivery
• Trace every number back to the tool call that produced it

• Agent Orchestrator — Understand the AI's decision-making in depth
• Charts & Data Visualization — Chart types and options
• Query Execution — How Trino queries run
• Skills & Marketplace — Install SQL and optimizer skills
• File Output — Export formats and viewer capabilities