Understanding AI Analysis

Starting with the rebuttal speeches, DebateFlow's AI begins helping you track what's happening in the debate. Think of it as having a second set of eyes—one that never gets confused about which argument is which.

When AI analysis appears

You won't see any AI analysis during the first two speeches (the constructive cases). That's intentional—we want you to form your own first impressions without any AI influence.

AI analysis starts appearing with the second affirmative speaker (the first rebuttal). From that point on, you'll see two new columns on your screen:

• "What Just Happened" - A summary of the previous speech
• Argument Tracker - Arguments identified throughout the round

"What Just Happened"

After each speech (starting with rebuttals), you'll see a "What Just Happened" column with our AI's analysis of what the previous speaker just said.

This gives you a quick sense of:

• What arguments were made
• How they structured their speech
• What they're claiming happened in the round

See full details

You can click "See full details" on any argument to see its structure broken down:

• Claim - What are they arguing?
• Warrant - Why should you believe it?
• Impact - Why does it matter?

This helps you evaluate whether an argument is complete and well-structured. Click "Hide details" when you're done.

Argument Tracker

The Argument Tracker shows all the arguments (or contentions) that have been identified in the round so far.

You'll see:

• Which team made each argument
• What the argument is about
• Its current status (we'll explain this more in a moment)

You can click on any argument to see the same claim/warrant/impact breakdown. This is helpful when you're trying to remember what a specific contention was about.

Coverage analysis (starting with second rebuttal)

Once you get to the second negative speaker (the second rebuttal), the "What Just Happened" column starts showing even more detail.

Did they respond effectively?

You'll see coverage analysis that helps you understand whether the previous speaker responded effectively to their opponent's arguments.

This isn't telling you who's winning—it's just tracking whether arguments were addressed or ignored. You still make all the decisions about quality and persuasiveness.

Drop detection

The analysis will also flag if the speaker dropped (didn't respond to) specific arguments from the other team.

Again, this is just information. Sometimes debaters make strategic choices about what to respond to. But it's helpful to have a clear picture of what was and wasn't addressed.

Argument status tracking

As the round progresses, you'll start seeing more detail in the Argument Tracker about the status of each argument:

• Contested - Both teams are actively debating this
• Dropped - One team stopped responding to it
• Frontlined - The team is defending it against attacks

This helps you see which arguments are still "live" in the round and which ones have been abandoned.

Extension analysis (by final focus)

By the time you get to Final Focus speeches, the AI analysis is showing you the complete picture.

Which arguments carried through?

You'll see extension analysis showing how many arguments were extended through each stage of the round.

This matters because in Public Forum debate, arguments need to be extended through Summary into Final Focus to count. The AI helps you track whether teams followed this rule.

The Argument Flow Diagram

One of the most powerful visualizations in DebateFlow is the Argument Flow diagram—a visual map showing how every argument moved through the round.

What the flow diagram shows:

• All arguments - Every contention from both teams, organized by which team introduced it
• Speech progression - Columns for each stage (Constructive → Rebuttal → Summary → Final Focus)
• Argument status - Visual indicators showing if arguments were extended, dropped, or responded to
• Connection lines - Lines showing how arguments flow from one speech to the next

How to read the diagram:

Each argument starts in the Constructive column and flows rightward through the round:

• Solid lines - Argument was extended by the team in that speech
• Dashed lines - Argument was mentioned or responded to by opponents
• Missing lines - Argument was dropped (not mentioned in that speech)
• Color coding - Different colors for Aff and Neg arguments

When to use the flow diagram:

The flow is especially helpful when:

• Checking extension rules - Did teams extend arguments through Summary into Final Focus?
• Identifying drops - Which arguments were abandoned and when?
• Understanding clash - Where did teams actually engage with each other's arguments?
• Making your decision - Which arguments survived the entire round intact?

Where to find it:

During the round, you'll see a simplified version in the Argument Tracker panel. After the round, the complete flow diagram is available in the detailed analysis view (click "View Details" on your dashboard).

Post-round: Complete technical analysis

After the round ends, there's one more analysis step that happens.

When you finish the last speech, you might see a loading screen that says we're completing the analysis. This usually takes a few minutes.

During this step, we're doing an in-depth technical analysis:

• Reviewing transcripts from all stages
• Cataloguing which arguments were extended, dropped, or refuted
• Analyzing the quality of argument structure
• Identifying potential voting issues

This comprehensive analysis helps you write your decision and feedback.

Viewing the detailed analysis

After the round, you can click "View Details" to see a complete analysis page.

This page includes:

• Round Overview - Summary statistics
• Complete Transcript - Everything that was said
• All Your Notes - From throughout the round
• Round Trajectory - How momentum shifted
• Round Timeline - Speaker performance over time
• Argument Flow - Visual diagram of how arguments moved through the round
• Voting Issues - AI-identified key decision points
• Argument Quality Analysis - Evaluation of claim/warrant/impact completeness
• Speaker Performance - Individual metrics for each debater

This is helpful for writing your RFD and providing detailed feedback to teams.

Privacy & how it works

You might be wondering: how does this all work, and what happens to student data?

The technical side

DebateFlow uses Claude 3.5 Sonnet (an AI from Anthropic) to analyze the transcripts. We send the text of speeches along with your notes and reactions.

What we DON'T send:

• Audio recordings (deleted immediately after transcription)
• Student names (replaced with anonymous tokens like "Team A Speaker 1")
• School names (also anonymized)

Everything is encrypted and processed privately. The AI never "learns" from your data—each round is analyzed independently.

Why it helps

You might think: "I'm the judge, why do I need AI help?"

Here's the thing—debate rounds move fast. Really fast. Even experienced judges can miss when an argument gets dropped or struggle to remember which speech a contention first appeared in.

The AI doesn't replace your judgment. It just helps you track the technical details so you can focus on evaluating quality, weighing impacts, and making the final decision.

Think of it like having a really good flow—except you don't need to learn technical debate flowing to get the benefit.

What's next?

Now that you understand how AI analysis works, learn about:

Next Steps:

• Decision Wizard - Use the Decision Wizard to organize your thoughts and write your RFD
• Generating Feedback - Create helpful feedback using the 3 C's method
• Tabroom Integration - Submit your ballot to Tabroom seamlessly

Related Resources:

• Recording Your Round - How AI analysis integrates with the recording process
• Workflow Overview - See how AI analysis fits into the complete judging workflow

Help & Support:

• Quick Reference - Fast lookup for argument status indicators
• FAQ - Common questions about AI in judging
• Troubleshooting - Fix AI analysis or loading issues