AI can run for days on its own now

2026-06-24 · Source: Matthew Berman · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Robotics & Autonomous Systems · Depth: Intermediate, quick

Summary

An AI named Codeex autonomously cloned a full-featured parody of Microsoft Excel after being given a six-word prompt: "/G goal clone Excel fullfeature parody." The AI operated for over 12 days before being manually stopped, demonstrating its ability to open the real Excel desktop application, analyze its features, and replicate them. The resulting application is highly functional, featuring adjustable and highlightable columns, working formulas like "1+2", and comprehensive formatting capabilities, including an ascending sort function. This extended autonomous operation highlights the AI's capacity for complex, multi-day task execution and feature replication based on a high-level goal.

Key takeaway

For AI Engineers exploring autonomous agent capabilities, this demonstration shows that current models can sustain complex, multi-day tasks like software cloning from minimal prompts. You should consider designing agents with longer operational windows and higher-level goals, moving beyond single-shot interactions. This shifts the focus from immediate output to persistent, goal-driven execution, potentially accelerating development of feature-rich applications through AI-driven replication.

Key insights

An AI autonomously replicated complex software features for over 12 days from a concise prompt.

Principles

• AI can execute multi-day, complex cloning tasks.
• Concise prompts can drive extensive autonomous operation.

Method

Codeex opened the real Excel desktop app, analyzed every feature, and then cloned them into its own functional version.

In practice

• Automate UI feature replication.
• Develop functional software parodies.

Topics

• Autonomous AI Agents
• Software Cloning
• Microsoft Excel Parody
• Codeex AI
• Long-running AI Tasks
• Feature Replication

Best for: CTO, VP of Engineering/Data, Director of AI/ML, Prompt Engineer, AI Engineer, Software Engineer

Related on AIssential

• I let Codex run for 6 hours. Here’s what happened. — Codeex's "goals" feature enables autonomous, multi-hour AI task execution by defining measurable outcomes and iterative verification.
• 5 Claude Code Agents You Can Build in less than 10 Minutes — Isolated, single-file AI agents can automate repetitive development tasks, reducing token burn and mental bandwidth.
• Long-running Agents — Long-running agents overcome AI limitations by externalizing state, enabling sustained, multi-session progress on complex tasks.
• You Are the Pilot Now — AI agents autonomously execute complex, multi-step tasks by planning, using tools, and adapting, unlike simple chatbots.
• Claude Agents Just Built a Fully Functioning C Compiler With Zero Human Input — AI agents can autonomously build complex software systems through coordinated, iterative development.
• Your First AI Agent — How to Build Autonomous Workflows That Work While You Sleep — Prompt to… — AI agents autonomously pursue multi-step goals, making decisions and using tools, unlike single-query prompts.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Matthew Berman.