AI Session Intelligence

The Only Platform Tracking
Individual AI Coding Sessions

119 session-level metrics from your real coding sessions. Measure what other platforms can't: actual AI productivity and effectiveness.

$ npm install -g guidemode
$ guidemode setup

Installs session capture into your AI coding agent. Nothing to leave running.

What Other Platforms Miss

Other DX platforms track AI "adoption" (did someone use an AI tool?). GuideMode tracks AI effectiveness (did the AI help or hurt productivity?).

Other Platforms

  • ❌ "AI adoption rate" (binary: used or not)
  • ❌ No session-level tracking
  • ❌ No AI provider comparison
  • ❌ No productivity metrics
  • ❌ Can't answer: "Did AI make us faster?"

GuideMode

  • ✓ 119 session-level metrics
  • ✓ Full conversation tracking
  • ✓ Compare AI providers side by side
  • ✓ Plan mode & todo efficiency
  • ✓ Answers: "Which AI makes us most productive?"
100%
Opt-In, Per Repository
119
Session Metrics
7
Productivity Dimensions
6
Productivity Measures

One Format for Every Assistant

Each assistant writes its own session format. GuideMode converts them all to one canonical shape, so your analytics don't depend on which tool produced the session.

claude-code icon
Available now

Claude Code

Full session tracking with plan mode, todo lists, and git operations

codex icon
Available now

OpenAI Codex

Full transcript and tool-call tracking from Codex rollout sessions

Coming Soon
cursor icon
AI Provider

Cursor

Interaction monitoring and conversation tracking

Coming Soon
github-copilot icon
AI Provider

GitHub Copilot

Session analytics and code suggestion tracking

Coming Soon
gemini-code icon
AI Provider

Gemini Code Assist

Conversation and tool usage tracking

Coming Soon
opencode icon
AI Provider

OpenCode

Session monitoring and analytics

Compare AI providers side-by-side to see which makes your team most productive

119 Session-Level Metrics

Deep insights into every AI coding session — from conversation quality to code output

Conversation Metrics

Message count, turn-taking, context length, token usage, conversation depth

Plan Mode Tracking

Plan mode usage, plan quality, implementation success rate, plan-to-code ratio

Todo List Analysis

Todo creation rate, completion rate, task breakdown quality, progress tracking

Git Operations

Commits per session, file change count, git efficiency ratio, commit message quality

Code Quality

Lines of code, file operations, edit efficiency, code review readiness

Productivity Metrics

Session duration, tasks completed, output quality, developer satisfaction

Session Success

Goal achievement, error resolution, code acceptance rate, user satisfaction

Tool Usage

Tool call frequency, tool success rate, tool diversity, automation level

Learning Velocity

Improvement over time, skill development, pattern recognition, efficiency gains

AI Productivity Cube

Dedicated analytics cube for AI productivity tracking with 7 dimensions and 6 measures

Dimensions

• AI Provider
• User
• Team
• Project
• Time Period
• Session Type
• Task Category

Measures

• Session Count
• Avg Duration
• Success Rate
• Code Output
• Task Completion
• Satisfaction Score

Example Analysis: "Which AI provider has the highest success rate for refactoring tasks on the backend team?"

AI Session Analytics

Real Questions, Real Answers

AI session intelligence lets you answer questions other platforms can't

"Which AI provider makes our team most productive?"

Compare session success rates, task completion, and code output across Claude Code, Cursor, Copilot, Gemini, and OpenCode

"Is our AI productivity improving over time?"

Track learning velocity, session efficiency trends, and skill development month over month

"Does plan mode actually help?"

Compare success rates for sessions with vs without plan mode across different task types

"What's our actual ROI on AI coding tools?"

Measure productivity gains, time savings, and code quality improvements with hard data

"Which developers need AI coaching?"

Identify struggling developers and share best practices from high performers

"Are we using AI for the right tasks?"

Break down task categories by success rate to find AI's sweet spot

Opt-In, Repository by Repository

Nothing is captured until you say so, and you say so per repository. Analysis runs on the server, so sessions are redacted on your machine before they ever leave it.

Redacted Locally

Secrets, tokens and your home directory are stripped before upload

Per-Repository Control

Allow or deny each repository, or deny everything by default

Off Whenever You Want

guidemode disable stops it here, or everywhere

Two Commands, Then Forget About It

The GuideMode CLI installs hooks into your coding assistant. There is no app to launch and no background process to babysit — capture happens as you work.

$ npm install -g guidemode
$ guidemode setup

Installs session capture into your AI coding agent. Nothing to leave running.

1

Sign in

Setup opens your browser and stores an API key locally.

2

Hooks installed

Claude Code and Codex hooks, plus a Claude Code skill so you can control syncing by just asking.

3

Choose what syncs

Pick which repositories are captured, or start with--default-policy deny and opt in one at a time.

Prefer not to install anything globally? npx guidemode runs the same first-run flow. Every session is hashed before upload, so an unchanged transcript transfers nothing at all.

Start Tracking AI Productivity

The only platform measuring actual AI effectiveness, not just adoption. Get insights other platforms can't provide.