Open-SWE: Opensource Jules! FULLY FREE Async AI Coder IS INSANELY GOOD!

Key Concepts

• Asynchronous coding agent
• Open Sway (Langchain)
• Jules (Google)
• Langraph
• Open-source alternative
• Local deployment
• API integration (Open Router, Olama)
• GitHub integration
• Codebase understanding
• Execution planning
• Code changes
• Pull requests
• Environment variables
• GitHub App creation
• Langraph agent
• Web application

1. Introduction: Jules by Google and the Need for Alternatives

• The video begins by discussing Jules, Google's asynchronous coding agent, highlighting its capabilities in reading code, writing tests, and fixing bugs autonomously.
• Jules operates asynchronously, allowing developers to continue coding while tasks are handled in the background.
• However, the video points out that Google has introduced a subscription plan for Jules, placing many features behind a paywall.
• Additionally, Jules primarily uses Gemini 2.5 Pro, which may not be the optimal choice for coding compared to newer models like Claude Opus 4.1 or GPT-5.
• This leads to the introduction of Open Sway by Langchain as a fully free, open-source alternative to Jules.

2. Open Sway: A Free and Open-Source Alternative

• Open Sway is presented as an asynchronous coding agent built with Langraph, designed to replicate Jules' functionality.
• It can autonomously understand codebases, plan solutions, and execute code changes across repositories, from initial planning to opening pull requests.
• A key advantage of Open Sway is its local availability, allowing users to bring their own APIs.
• For users without API funds, Open Router (free provider) or Olama (local model hosting) are suggested as alternatives.

3. Functionality and Features of Open Sway

• Open Sway connects directly with GitHub repositories and can take tasks from issues or a custom UI.
• It researches the codebase, creates execution plans, writes and tests code, reviews its own work, and opens pull requests, mirroring Jules' workflow.
• Users can access Open Sway through a cloud server or install it locally.

4. Demo: Creating a Calendar App with Open Sway

• The video showcases a demo of Open Sway creating a calendar app.
• The agent creates a sandbox, clones the repository, and checks out the branch.
• It uses planning and execution agents to execute the task, showing reasoning and the tools used.
• The agent generates a script, and upon user approval, executes the task, planning progress across multiple sub-tasks.
• Open Sway demonstrates impressive intelligence by recognizing an existing pull request for the same task and building upon it.
• The user can review the task list, track progress, and see summaries of completed tasks.
• Token usage, input/output tokens, and costs are displayed, even highlighting potential savings through caching.
• After completing tasks, Open Sway analyzes the code and implements changes or fixes errors.
• The demo concludes with a basic calendar app that allows adding, editing, and deleting events.

5. Local Installation of Open Sway: Prerequisites and Steps

• The video provides a step-by-step guide to installing Open Sway locally.
• Prerequisites:
  • Git (for cloning the repository)
  • IDE (e.g., VS Code, Curtis, Windsurf)
  • Free API key from Open Router or local model hosting (Olama)
• Installation Steps:
  1. Clone the repository using the git clone command.
  2. Navigate to the cloned directory using cd open-sway.
  3. Install dependencies using the yarn command.
  4. Open the project in an IDE.
  5. Set up environment files by copying the example files for both the web app and the agent.
  6. Configure the web app environment variables (Langraph API, GitHub app client).
  7. Configure the agent environment variables (LM providers, tools like firecrawl, local model provider).
  8. Create a GitHub app, filling out basic information, configuring OAuth settings, and setting up a webhook (using Enro if needed).
  9. Collect information for the required variables and paste them into the environment variable file.
  10. Start the Langraph agent using the yarn dev command.
  11. Start the web application using a terminal tool.
  12. Access the application via localhost.

6. Environment Variables and GitHub App Configuration

• The video emphasizes the importance of configuring environment variables for both the web app and the agent.
• The web app environment variables include the Langraph API, the next public API, and the GitHub app client.
• The agent environment variables allow setting the LM providers, tools like firecrawl, and infrastructure configurations.
• Creating a GitHub app involves filling out basic information, configuring OAuth settings, and setting up a webhook.
• Enro can be used to set up the webhook, and the generated URL should be saved as a GitHub web hub secret.

7. Starting the Langraph Agent and Web Application

• After configuring the environment variables and creating the GitHub app, the video explains how to start the Langraph agent and the web application.
• The yarn dev command is used to start the Langraph agent.
• A terminal tool is used to start the web application.
• Once both are running, the application can be accessed via localhost.

8. Conclusion

• Open Sway is presented as a compelling open-source alternative to Jules, offering similar features and local deployment options.
• The video encourages viewers to explore Open Sway and provides links to resources in the description.
• The video concludes with a call to action, encouraging viewers to subscribe to the channel, join the newsletter and Discord, and follow the presenter on Twitter.

9. Notable Quotes

• "Open sway is an asynchronous coding agent built with Langraph. It can autonomously understand your entire codebase, plan solutions, and execute code changes across your repository from initial planning all the way to opening pull request."
• "And the best part is it's locally available which lets you bring your own API. And if you don't have the funds for an API, you can either use open router as a free provider or you can use Olama where you can locally host these models without paying for any API costs."

10. Technical Terms and Concepts

• Asynchronous Coding Agent: A tool that automates coding tasks in the background, allowing developers to continue working without interruption.
• Langraph: A framework for building conversational AI applications.
• Open Router: A platform that provides access to various AI models through a single API.
• Olama: A tool for running language models locally.
• GitHub App: An application that integrates with GitHub to automate tasks and extend functionality.
• Webhook: A mechanism for receiving real-time updates from GitHub.
• Environment Variables: Variables that store configuration settings for an application.
• LM Providers: Providers of language models, such as OpenAI, Anthropic, or local models.

11. Synthesis/Conclusion

The video effectively presents Open Sway as a viable and free alternative to Google's Jules, particularly for developers seeking an open-source, locally deployable solution. It provides a detailed overview of Open Sway's features, a practical demonstration of its capabilities, and a comprehensive guide to local installation, empowering viewers to adopt and utilize this tool in their own development workflows. The emphasis on cost-effectiveness through local model hosting and free API options further enhances its appeal.