Discuss with AI Chat Feature Guide

Learn how to analyze your PB Vision matches using AI assistants such as ChatGPT, Claude, Gemini, and other MCP-compatible platforms. This guide covers both downloadable AI chat files and direct MCP integrations for advanced workflows.

📍 Quick Start Summary

PB Vision offers two ways to analyze your matches further with your favorite AI chat agents

📥 Download Chat

  • Works with ChatGPT, Claude, Gemini, and most AI assistants

  • No setup required

  • Download a chat file and upload it to your preferred AI tool

🔗 MCP Connection

  • Connect AI assistants directly to PB Vision

  • No manual downloads required

  • Best for users familiar with MCP-compatible platforms

Feature

Download

MCP

Easy Setup

✅

❌

Works with ChatGPT

✅

✅

Works with Claude

✅

✅

Requires Configuration

❌

✅

Recommended for Most Users

✅

❌

🌱 Download Chat File

This works with any AI chatbot, free or paid, including ChatGPT, Claude, Gemini, and others.

How It Works

  • Open any processed game in PB Vision and select Discuss with AI Chat.

  • Download the generated data file.

  • Open your preferred AI assistant, such as ChatGPT, Claude, or Gemini.

  • Upload the downloaded file and ask a question about your match.

📝 Recommended Prompt

I am [Player Name]. Analyze this match and identify my strengths, weaknesses, and areas for improvement.

The AI will use your PB Vision data to provide personalized insights and answer questions about your performance.

Single Game

Generate a chat file for an individual match. This option includes match statistics, player ratings, shot analysis, and performance insights for a single game.

To get started, open the bubble-dot menu from either your Video Library or an individual video and select Discuss with AI Chat.

From Video Library

From Video

Multi-Game

Generate a chat file using multiple matches to identify longer-term trends and patterns. This option helps AI assistants provide broader feedback on consistency, strengths, weaknesses, and areas for improvement across several games.

To get started, select multiple videos/folders from the video library and choose Discuss with AI from the available actions.

Tips for Best Results

Use File Uploads

  • Attach the downloaded chat file directly to your AI assistant instead of copying and pasting the contents. Large match files may exceed text limits when pasted manually.

Ask Focused Questions

  • For longer matches, consider asking about specific areas such as serves, returns, third shots, or unforced errors. Focused questions often produce more detailed insights.

Note - Longer games can use more tokens from your AI chat agent.

Most AI Assistants Are Supported

  • The downloaded file can be used with ChatGPT, Claude, Gemini, and other AI assistants that support file uploads. Paid plans may support larger files and provide more comprehensive responses.

Tagged Player Names Are Included

  • If players are tagged in PB Vision, their names will be included in the exported data, allowing the AI to reference them throughout the analysis.

Video Walkthroughs

🧠 MCP Setup

MCP (Model Context Protocol) allows supported AI assistants to access your PB Vision game data directly. Instead of downloading and uploading chat files, you can connect your AI client to PB Vision and analyze matches on demand.

This option is best for users who want a more streamlined workflow and are comfortable configuring AI tools.

Requirements

  • An AI client that supports MCP servers or custom connectors (such as ChatGPT, Claude Desktop, Cursor, Gemini CLI, or Gemini Code Assist).

  • A PB Vision account with uploaded and processed games.

Connect to PB Vision

Note - If your client allows configuring the HTTP method, ensure it is set to POST.

No API keys or authentication are required. Simply copy the MCP URL below and add it to your supported AI platform.

MCP URL

https://share.pb.vision/mcp

Claude (Desktop)

Note - You'll need a paid Claude plan (Pro or Max).

Quick Steps

Open Claude Desktop → Select the “+” button → Connectors → Add a custom MCP connector → Paste the PB Vision MCP URL → Complete setup and begin chatting.

ChatGPT

🚧 Under Construction 🚧

More information on connecting MCP with ChatGPT will be added soon.

Gemini Advanced (CLI)

Note - Gemini CLI must be installed locally on your machine. You’ll also need a Google Gemini subscription that supports Code Assist features.

Connect PB Vision directly to Gemini CLI using MCP. Once connected, Gemini can access your PB Vision match data and analyze games without requiring file downloads.

Quick Steps

Open the Gemini settings file → Add the PB Vision MCP server configuration → Launch Gemini CLI → Verify the connection → Start analyzing your PB Vision data.

Gemini Code Assist (VS Code)

Note- Gemini Code Assist must be installed and enabled in VS Code, and you must be signed in to a Google Gemini subscription that supports Code Assist features.

Connect PB Vision directly to Gemini Code Assist within Visual Studio Code. Once connected, Gemini can access your PB Vision match data and answer questions without requiring file downloads.

Quick Steps

Open your Gemini settings file → Add the PB Vision MCP server configuration → Reload VS Code → Enable Agent Mode → Verify the connection → Start analyzing your PB Vision data

🎯 Using PB Vision with MCP

Once connected, your AI assistant gains access to several PB Vision tools that can retrieve match data, identify players, and analyze performance.

Available MCP Tools

Once connected, your AI assistant gets access to five tools:

Tool

What It Does

Best Used For

get_analysis_guide

Loads PB Vision’s analysis
framework and scoring guidelines.

Starting a new
analysis session

get_video_insights

Retrieves detailed match data including
shots, outcomes, and performance
metrics.

Deep analysis of
a specific match

get_insights_schema

Explains the meaning of PB Vision
statistics and metrics.

Understanding
advanced stats

get_player_images

Retrieves player snapshots to help
identify participants in a match.

Confirming player
identity

get_download_url

Generates a downloadable version
of match data.

Long matches and
token-efficient
analysis.

How to Use MCP

Find Your Video ID

  • Your video ID is the 12-character code found in your game URL.

Example Game URL

https://pb.vision/video/ox31i0pijbzo/0/overview

Video ID

ox31i0pijbzo

Example Prompts

  • Break down my play from game ox31i0pijbzo

  • Analyze my serve performance in game ox31i0pijbzo

  • Compare my stats across games ox31i0pijbzo and 93vkk5bap5he

The AI will use PB Vision tools to retrieve match data and provide analysis.

MCP Tips

Start with the Analysis Guide

  • If the AI seems unsure how to interpret your data, ask it to run the analysis guide first.

Multi-Game Analysis

  • Provide multiple video IDs to compare performance across games.

Large Games

  • For longer matches, use download-based analysis tools when available to reduce token usage.

MCP Troubleshooting

Connection Failed or No Tools Appear

  • Verify that your MCP client is configured to use POST requests instead of GET. If you’ve recently added the PB Vision MCP server, restart your AI client and try again.

AI Cannot Access Match Data

  • Confirm that you’re using the correct 12-character video ID from your PB Vision game URL.

Example Game URL

https://pb.vision/video/ox31i0pijbzo/0/overview

Video ID

ox31i0pijbzo


Responses Are Incomplete or Cut Off

  • Long matches can generate large amounts of data and may exceed your AI platform’s context limits.

    For longer games:

    • Ask the AI to use get_download_url

    • Focus on specific aspects of the match (serves, returns, drops, etc.)

    • Analyze one game at a time

⁉️ Need Help?

This feature is in beta and we're actively improving it. If you run into issues or have feedback: