Listen to this story
|
Devin, the world’s first AI engineer by Cognition Labs, took the internet by storm with its ability to write code from scratch, fix bugs, and deploy solutions, aiming to automate aspects of the software development process. However, Devin is not the only AI autonomous agent available
Here is the list of top Devin alternatives.
Top Devin AI Alternatives for Coding Automation in 2024
Name | Developed by | Price |
Devika AI | Mufeed VH | Free |
Replit Code Repair | Amjad Masad | Free |
SWE Agent | John Yang | Free |
OpenDevin | community-driven project | Open Source |
MetaGPT | Pico | Open Source |
ChatDev | community-driven project | Open Source |
1. Devika
Devika is an open-source AI software engineer created by Mufeed VH (Hamzakutty), the founder of Lyminal and Stition.AI. It is capable of understanding human instructions, breaking them down into tasks, conducting research, and autonomously writing code to achieve set objectives.
Devika aims to be a competitive open-source alternative to Devin by Cognition AI. It utilises LLMS, planning and reasoning algorithms, and web browsing abilities to intelligently develop software.
One of Devika’s key strengths lies in its ability to function as an AI pair programmer, reducing the need for extensive human intervention in complex coding tasks.
Devika simplifies software development processes, whether it’s creating new features, debugging code, or developing entire projects from scratch, thereby enhancing efficiency.
The main difference between Devin and Devika, apart from the latter being open source, is that Mufeed used Claude 3 instead of GPT-4 for Devika.
2. Replit Code Repair
Replit’s Code Repair is a low-latency code repair AI agent. It utilises LLMs trained on a massive dataset of code examples and their corresponding fixes. This allows the LLM to analyse your code and identify potential errors or inefficiencies.
Replit took a 7B Code LLM and fine tuned it into a tool that mimics the behavior of LSP Code Actions. The special ingredient is in the training data — a careful mixture of real-world errors (collected on Replit) combined with synthetically-generated code fixes.
Replit’s approach involves using Operational Transformations (OTs) and session events to create a dataset of (code, diagnostic) pairs. They synthesise diffs using large pretrained code models and fine-tune them for code repair tasks.
Explore: Latest News & Stories About Replit
3. SWE Agent
SWE Agent is also an open-source alternative to Devin, much like Devika, developed by a team led by John Yang, Carlos E. Jimenez, and Alexander Wettig at Princeton University. It turns language models like GPT-4 into software engineering agents that can fix bugs and issues in actual GitHub repositories.
On the full SWE-bench test set, SWE-Agent resolves 12.29% of issues. The key to SWE-Agent’s success lies in its innovative Agent-Computer Interface (ACI), which streamlines the interaction between the language model and the code repository.
Unlike traditional approaches, SWE-Agent’s ACI simplifies commands and feedback formats, making it easier for the model to navigate, edit, and execute code files within the repository. Developers can easily set it up using Docker and Miniconda, following straightforward installation and configuration steps outlined in the project’s documentation.
4. OpenDevin
OpenDevin is an open-source project aiming to mimic Devin, an AI software engineer. Similar to Devin, OpenDevin aspires to handle various aspects of software development, potentially including, Code Generation, Debugging, and Deployment Automation
The alpha version is available for testing, showcasing its ability to handle complex tasks and collaborate with users.
The project is focusing on key milestones creating a user-friendly interface with chat and command features, building a stable backend for commands, improving the agent’s capabilities, and setting up an evaluation pipeline.
Explore: Latest News & Stories About Devin
5. MetaGPT
MetaGPT is a multi-agent framework that in itself acts as a virtual software company. It takes a one-line requirement and outputs user stories, competitive analysis, requirements, data structures, APIs, and documents.
MetaGPT includes product managers, architects, project managers, and engineers, following carefully crafted Standard Operating Procedures (SOPs).
Explore: Latest News & Stories About Meta
6. ChatDev
Similar to MetaGPT, ChatDev stands as a virtual software company that operates through various intelligent agents holding different roles, including Chief Executive Officer , Chief Product Officer , Chief Technology Officer , programmer , reviewer , tester , art designer.
These agents form a multi-agent organisational structure and are united by a mission to “revolutionise the digital world through programming.”They collaborate through specialised functional workshops at ChatDev, engaging in activities like design, coding, testing, and documentation.