Back to Lab
AI Agent IDE

Cursor

Cursor is our daily-driver AI editor for precise coding, debugging, refactors, and working with complex codebases at speed.

Stats

2025 by the numbers

Tokens
2,729,456,396

processed

Agents
5,200

launched

Messages
5,229

sent

Days
85

active

Our year with Cursor

Proof & Track Record

Click to info
2025 - Milestones and Evolution

2025 - Milestones and Evolution

2025 - Milestones and Evolution

2025 was full of innovations: 16 versions, 26 models, Bugbot, Cloud Agents, and Cursor 2.0 with Composer 1. Key milestone: May 22nd - Anthropic released Claude Opus 4, which changed the way we work with AI coding.

Return
Click to info
2.73 Billion Tokens Used

2.73 Billion Tokens Used

2.73 Billion Tokens Used

Over 2.7 billion tokens were used during 2025. The most intensive period was from week 28, when massive work with AI agents began. Peak consumption was in week 50 with 823 million tokens - proof of growing productivity and project complexity.

Return
Click to info
Experimentation with 19 AI Models

Experimentation with 19 AI Models

Experimentation with 19 AI Models

During the year we tested 19 different AI models and main preferences changed 8 times. Composer 1 dominated from September, Claude 4.5 Opus experienced a renaissance in December, while GPT 5.1 Codex remained a stable complement. This experimentation allowed us to find optimal combinations for different types of tasks.

Return
Click to info
Evolution of AI Model Usage

Evolution of AI Model Usage

Evolution of AI Model Usage

The graph shows the dynamics of using different models from May to December. Gemini 2.5 Pro dominated in summer, Composer 1 shone in autumn, GPT 5 had its peaks in August and October. This model diversity ensures optimal performance for each type of task.

Return
Click to info
Most Used Models

Most Used Models

Most Used Models

Out of 19 tested models, 5 favorites emerged: Gemini 2.5 Pro (1,651 prompts), Composer 1 (1,553 prompts), GPT 5 (975 prompts), GPT 5.1 Codex (350 prompts), and Claude 4.5 Opus (330 prompts). These models form the backbone of our productivity.

Return
Click to info
5,229 Messages with AI Agents

5,229 Messages with AI Agents

5,229 Messages with AI Agents

During the year we sent 5,229 messages in 909 chats with AI agents. Activity started in July (1,202 messages) and peaked in December (1,591 messages). This growth shows how AI agents have become an integral part of our workflow.

Return
Click to info
85 Days of Active Use

85 Days of Active Use

85 Days of Active Use

Cursor was actively used 85 days during 2025, with the longest streak of 9 days in July. The activity calendar shows consistent work from May, with increasing intensity towards the end of the year. Each day of use meant more productive development.

Return
Click to info
Most Active Periods

Most Active Periods

Most Active Periods

We used Cursor most in July (month peak), on Tuesdays (most productive day of the week), and between 22:00-00:00 (evening productive window). This data helps optimize workflow and plan the most demanding tasks for the most productive times.

Return
Click to info
Complete 2025 Overview

Complete 2025 Overview

Complete 2025 Overview

Year summary: 235 days since start of use, 5.2K agents launched, 4 active tabs, 2.73 billion tokens processed, and a 9-day streak. Top 3 models: Gemini 2.5 Pro, Composer 1, and GPT 5. These numbers prove that Cursor isn't just a tool, but a key part of our development process.

Return

About Cursor

What it is and why we use it

What is Cursor?

Cursor is an AI-first code editor built on VS Code that combines the power of AI models with a powerful editor. It enables faster development through intelligent suggestions, contextual code chatting, and advanced features for working with the entire codebase.

Why we use it?

We use Cursor for specific use cases where detailed code work, analysis, and debugging of complex parts of applications is needed. Ideal for refactoring, debugging, and working with legacy code where we need full control over the process.

Key Benefits

AI Code Completion - Intelligent suggestions while typing
Codebase Chat - Questions about the entire project in natural language
Multi-file Editing - Edits across multiple files at once
Codebase Indexing - Fast search across the entire project
VS Code Compatible - All VS Code extensions work
Terminal & Git - Integrated tools for development

Track Record

Results & Achievements

Legacy Code Refactoring

Modernization of older projects with AI analysis and automated modifications

50+ projects refactored, 80% time savings

Complex Bug Debugging

Rapid identification and fixing of bugs across the entire codebase thanks to AI analysis

95% solution success rate, average 3x faster

Code Review & Optimization

Automatic code quality checks and optimization suggestions

12,500+ files reviewed, 1,200+ optimizations

Documentation & Explanation

Generating documentation and explanations of complex code for the team

100% documentation coverage, 60% time savings

Year Summary

2025 with Cursor

May

Claude Opus 4 - Key Milestone

May 22nd - Anthropic released Claude Opus 4, which changed the way we work with AI coding

July

Most Active Month and Longest Streak

1,202 messages with agents, 9-day streak - productivity peak

September

Composer 1 Dominates

Composer 1 became the main model, experimentation with 19 models continues

November

Claude 4.5 Opus Renaissance

Claude 4.5 Opus experiences a comeback, GPT 5.1 Codex stable complement

December

Record Year Completed

2.73 billion tokens, 5,229 messages, 85 active days - a year full of growth

Game Changing Facts

Why Cursor isn't just an editor, but a competitive advantage

2.73 billion tokens

That's more than most developers use in their entire career. We did it in a year.

85 active days

Each day of use meant more productive development. Longest streak: 9 days in a row.

19 tested models

We experimented with every available model to find optimal combinations.

5,229 messages with agents

AI agents have become an integral part of our workflow. We can't work without them anymore.

95% solution success rate

Debugging complex bugs is on average 3x faster than without AI support.

80% time savings

Refactoring legacy code that would take weeks, we complete in days.

Your competition is already using Cursor

While you're thinking about it, others are gaining a lead. Every day of waiting means lost opportunities and competitive disadvantage. Top developers worldwide have already switched to AI-first development. The question isn't whether you'll start using Cursor, but when.

What you lose every day
  • Hours of productive time
  • Faster project delivery
  • Better code quality
  • Competitive advantage
What you get immediately
  • 3x faster debugging
  • 80% time savings on refactoring
  • AI agents on your side
  • Lead over competition