Introducing the Flarelight AI Assistant

We've redesigned the Flarelight home page around a conversational AI assistant — your starting point for everything in the platform. Instead of navigating menus to get things done, you can just ask.

---

What the AI Assistant can do

The assistant understands your workspace and can take real action on your behalf. Here's a breakdown of what it handles across each area of the platform.

Data Import

• Upload a file directly through the chat (CSV, JSON, XML, XLSX)
• Ask the AI to generate sample data from a plain-text description — it creates and imports the dataset automatically
• Navigate to your imports, refresh an existing import, or delete one

Data Preparation

Raw imports are rarely ready to use as-is. Data preparation is where you shape them into something useful — and the assistant can handle the heavy lifting.

• Combine multiple imports into a single dataset — merge sales data from different regions, join customer records with transaction history, or consolidate monthly exports into one clean table
• Add calculated or grouping columns — ask the AI to add a column that groups products into categories, flags high-value customers, or maps date ranges to quarters so your dashboards can filter and segment properly
• Generate a full preparation configuration from scratch — describe what you want the output to look like and the assistant analyses your sources and builds the data model automatically
• Run a preparation and wait for the result, all from the chat
• Navigate to any preparation or delete one you no longer need

Analysis

The assistant can create and run a wide range of statistical analyses by name. Supported types include:

Descriptive & exploratory

• Descriptive Statistics
• Correlation Analysis
• Box-Whisker Plot, Q-Q Plot

Normality tests

• Shapiro-Wilk, Kolmogorov-Smirnov, Anderson-Darling

Hypothesis tests

• One-sample and two-sample T-Tests and Z-Tests
• Paired T-Test and Z-Test
• Welch Two-Sample Test
• One-Way ANOVA, Welch ANOVA

Non-parametric tests

• Mann-Whitney U Test
• Kruskal-Wallis Test
• Wilcoxon Signed-Rank Test (one-sample and paired)

Just say "run a correlation analysis on my sales data" and the assistant picks the right type, creates the analysis, and applies an AI-generated configuration automatically.

Reporting

• Create a new report
• Ask the AI to generate a report configuration from your data sources — it sets up the layout, charts, and fields without you touching the editor
• Trigger report generation and wait for it to complete
• Open the report editor for manual refinements

Automation

• Create and configure automations through conversation
• Ask the AI to generate an automation configuration from a description of what you want it to do
• Trigger an automation run and track execution status

---

Workspace sidebar

Alongside the chat, the home page shows a live workspace snapshot: counts of your imports, preparations, analyses, reports, and automations, plus a recent activity feed so you always know where you left off.

Every item in the sidebar is a direct link. Click any resource in the activity feed or workspace summary to jump straight to it — no searching, no navigating through menus.

---

Resuming where you left off

The assistant remembers your recent activity. Open the app and it will greet you with what you last worked on and offer to pick up from there. Or just type "continue where I left off."

---

Mentioning resources in chat

You can @mention any existing import, preparation, analysis, or report directly in the chat to refer to it by name — without having to copy and paste IDs.

---

What the AI assistant cannot do

The assistant works at the application level — it creates, configures, and runs things on your behalf. It does not have access to the actual contents of your data. It can't read your files, search across datasets, or reason over the values inside a preparation or analysis result.

---

Feature and permission aware

The assistant only surfaces capabilities your subscription includes and your account has access to. If a feature isn't on your plan, it will say so rather than offering actions that won't work.

For full details on how AI is used and what data is sent, see our AI Policy.

---

Getting started

Open the Flarelight home page and start a conversation. Some things to try:

• "Import a CSV file"
• "Create a correlation analysis called Sales Trends"
• "Build a report from my Q1 preparation"
• "Set up an automation that runs every Monday"
• "What can you help me with?"