Skip to main content
RelayHub includes a suite of built-in data analysis tools that the AI uses automatically when you ask questions about uploaded spreadsheets, CSVs, or any structured data. You do not need to learn special commands — just ask questions in natural language and the AI selects the right tools, runs the analysis, and generates charts directly in the chat.

How It Works

When you upload a data file or ask analytical questions, the AI has access to 15 specialized tools for data processing. It automatically determines which tools to use based on your request. For example, asking “Show me the revenue trend over the last 12 months” triggers the AI to:
  1. Read the relevant data file
  2. Identify the date and revenue columns
  3. Run a time series analysis
  4. Generate a line chart
  5. Present the chart with a written summary of the trend
All of this happens in a single response — you ask a question and get back a chart with commentary.

Analysis Focus Modes

When the AI analyzes your data, it selects a focus mode based on what you are asking:
ModeWhat It DoesBest For
ComprehensiveFull analysis across multiple dimensions with several chart types”Give me an overview of this dataset”
DistributionHistograms, box plots, and statistical summaries of how data is spread”How is salary distributed across the company?”
CorrelationScatter plots and heatmaps showing relationships between variables”Is there a relationship between ad spend and conversions?”
Time SeriesLine charts and trend analysis over temporal data”Show me monthly revenue for the past year”
ComparisonBar charts and ranking visualizations for categorical data”Compare sales performance across regions”
You can steer the AI toward a specific focus by being explicit in your request. “Show me the distribution of response times” will produce histograms, while “Compare response times across teams” will produce bar charts.

Chart Types

The AI generates publication-ready charts directly in the conversation:

Bar & Stacked Bar

Categorical comparisons and composition breakdowns

Line

Trends over time and sequential data

Scatter

Relationships between two numeric variables

Histogram & Box

Data distribution and statistical spread

Pie

Proportional composition of a whole

Heatmap & Ranking

Correlation matrices and ordered comparisons

Explore and Compute Tools

Beyond full analysis, the AI can perform quick data operations: Explore Data lets the AI preview your dataset before running a full analysis. It supports actions like:
  • Preview — See the first few rows and column types
  • Describe — Get statistical summaries (mean, median, min, max, standard deviation) for every numeric column
  • Columns — List all columns with their data types
Compute Data runs specific calculations without generating charts:
  • Sum, Mean, Median — Aggregate a column
  • Group By — Break down metrics by category
  • Rank — Order rows by a specific column
  • Pivot — Restructure data into a pivot table
  • Correlation — Calculate correlation coefficients between columns

Inline Data Analysis

You do not need an uploaded file to generate charts. The AI can visualize data from any source that appears in the conversation, including:
  • Data returned from company knowledge base queries
  • Numbers you type directly into the chat
  • Tables extracted from documents discussed earlier in the conversation
For example, if you paste a small table of monthly sales figures into the chat and ask “Chart this as a bar graph”, the AI converts it to inline data and generates the visualization on the spot.

Querying Spreadsheets with Natural Language

Uploaded CSV and Excel files can be queried using plain English. Ask questions like:
  • “What is the total revenue for Q3?”
  • “Show me all rows where the status is ‘overdue’”
  • “What is the average deal size by sales rep?”
The AI translates your question into a structured query, runs it against the spreadsheet data, and returns the results as a formatted table or chart.
Natural language queries work on any CSV or Excel file in Ready status. The AI reads the column headers to understand the data structure, so descriptive column names produce better results than generic ones like “Column A”.

Tips for Better Analysis

Very large datasets (50,000+ rows) may produce simplified visualizations. If you need detailed analysis of a large file, ask the AI to focus on a specific subset — for example, “Analyze only the 2025 records” or “Show me the top 100 customers by revenue.”
  • Be specific about what you want to see — “Show revenue by quarter” gives better results than “Analyze this file”
  • Name your columns clearly — AI performance improves when spreadsheet headers are descriptive
  • Ask follow-up questions — After an initial analysis, drill deeper with questions like “Now break that down by region” or “Remove the outliers and re-chart”
  • Request specific chart types — If you want a scatter plot instead of the bar chart the AI chose, just ask: “Show that as a scatter plot instead”