Tabloy AI Docs

English

Español
Français
العربية

English

Español
Français
العربية

Appearance

Using Tabloy AI ​

This page explains the full Tabloy AI flow from data connection to final dashboard sharing.

It combines:

  • product behavior shown on the marketing site,
  • real user flows in the desktop app (tabloy-native),
  • practical guidance for everyday usage.

The Core Idea ​

Tabloy AI is designed as a local-first analytics workspace:

  • your working data stays in the desktop runtime,
  • AI helps you plan and build analytics output,
  • you can share only what you explicitly publish.

In practice, this means you can move from raw data to a usable dashboard without changing tools.

End-to-End System Flow ​

%%{init: {"flowchart": {"curve":"linear","nodeSpacing":30,"rankSpacing":34}} }%%
flowchart TB
  A["💾 Data Source"] --> B["🖥️ Desktop Runtime"]
  B --> C["🤖 Tabloy Agent"]
  C --> D["📊 Build Dashboard"]
  D --> E["🔎 Explore & Refine"]
  E --> F["🔗 Share / Export"]

Source to Dashboard Journey ​

Step A: Connect your source ​

You start with either:

  • local files (CSV, JSON, XLSX),
  • SQL integrations (currently PostgreSQL and MySQL/MariaDB in desktop UI).

After connection/upload, select the source as Active.

Step B: Build understanding context ​

Tabloy prepares the source for analysis:

  • table and column structure is collected,
  • dataset context is prepared for AI-assisted planning.

Step C: Build with AI ​

In Build Mode, you describe what you want (for example, trends, KPIs, category breakdowns). Tabloy suggests and generates widgets and page structure.

Step D: Review and refine ​

You validate output:

  • check values and labels,
  • refine prompts,
  • adjust page structure and widgets.

Step E: Publish or export ​

When ready, you can:

  • publish a read-only snapshot link,
  • export results for reports and presentations.

Build Mode: How Decisions Are Made ​

%%{init: {"flowchart": {"curve":"linear","nodeSpacing":22,"rankSpacing":24}} }%%
flowchart TD
  A["✍️ Request"] --> B["🤖 Tabloy Agent"]
  B --> C{"🎛️ Mode"}
  C -->|Agent| D["⚙️ Update Widgets"]
  C -->|Chat| E["💬 Explain"]
  C -->|Auto| F["✨ Guided Start"]
  D --> G["👀 Preview"]
  E --> G
  F --> G
  G --> H{"✅ Quality"}
  H -->|Good| I["📌 Keep"]
  H -->|Needs tuning| J["🔁 Refine Prompt"]
  J --> A

What Each Mode Is Best For ​

Agent ​

Use when you want Tabloy to directly change dashboard output:

  • add chart/KPI/table,
  • update existing widgets,
  • evolve page content quickly.

Chat ​

Use when you want analysis-first help:

  • ask questions,
  • validate interpretation,
  • brainstorm metrics before changing layout.

Auto ​

Use for fast starts when you have a broad goal but no structure yet.

Local-First Data Model in Plain Language ​

%%{init: {"flowchart": {"curve":"linear","nodeSpacing":24,"rankSpacing":28}} }%%
flowchart LR
  subgraph L["🏠 Local Environment"]
    A["💾 Dataset"] --> B["🖥️ Runtime"] --> C["📊 Dashboard"]
  end
  B --> D["🤖 Tabloy Agent Context"]
  D --> C
  C --> E["🔗 Optional Share"]

Why this matters ​

  • You keep control over your working dataset.
  • You decide when to share outputs.
  • You can operate with both file-based and integrated SQL sources.

Explore Mode: Quality Control Layer ​

Explore is where you validate output quality before sharing:

  • inspect widget output,
  • verify whether charts answer the original question,
  • improve weak or ambiguous results with targeted follow-up requests.

Use Explore especially for high-visibility dashboards (weekly leadership reports, client updates, finance summaries).

Multi-Page Dashboard Workflow ​

For non-trivial reports, keep structure intentional:

  • Page 1: Executive overview,
  • Page 2: Trend analysis,
  • Page 3: Segment deep dive.

This mirrors how decision-makers consume information and makes snapshots easier to share.

Practical Prompt Framework ​

A reliable prompt usually includes:

  • Metric (what to measure)
  • Timeframe (for when)
  • Grouping (by what dimension)
  • Output preference (chart/KPI/table)

Example:

Show monthly revenue and profit for 2025 by region, then add a KPI summary card.

Sharing Model ​

Tabloy sharing is designed for controlled collaboration:

  • publish read-only snapshot links,
  • update snapshot after major dashboard edits,
  • keep internal context in your message (period, assumptions, target audience).

Typical "Good" Operating Pattern ​

  • Upload or connect source
  • Set source as Active
  • Build first draft with Agent mode
  • Validate key widgets in Explore
  • Organize pages
  • Publish snapshot and/or export
  • Iterate from feedback

Summary ​

Tabloy AI works as a loop:

  • connect source → build with AI → validate → share/export → iterate.

The strength of the workflow is not just generation speed, but the combination of:

  • local-first execution,
  • structured AI assistance,
  • controlled sharing for real team decisions.

Copyright (c) Tabloy