The bird-themed social network Twitter’s identity may have been X-ed out by new owner Elon Musk, but that isn’t stopped one of its co-founders, Jack Dorsey, from taking on new bird names for a new project.
Dorsey’s other company Block, the parent of point-of-sale service Square, mobile payments system Cash App, music streaming service Tidal, and other tech-driven financial tools, today launched Goose, a free, open-source framework that seeks to simplify the process of building an AI agent (or many agents) with pretty much any conceivable large language model (LLM) as the intelligence on the backend, whether that be DeepSeek or a proprietary model from the likes of OpenAI, Google, and Anthropic.
Less than 24 hours ago, Dorsey teased the launch of Goose in effect by posting on X: “open source everything.”
The agents that Goose is designed to build are primarily focused on software development, and in a blog post announcing the new framework, Block developer advocate Adewale Abati wrote that the framework has already helped with the following tasks:
- Conduct code migrations such as Ember to React, Ruby to Kotlin, Prefect-1 to Prefect-2 etc.
- Dive into a new project in an unfamiliar coding language
- Transition a code-base from field-based injection to constructor-based injection in a dependency injection framework.
- Conduct performance benchmarks for a build command using a build automation tool
- Increasing code coverage above a specific threshold
- Scaffolding an API for data retention
- Creating Datadog monitors
- Removing or adding feature flags etc.
- Generating unit tests for a feature
The announcement was made today by Block’s Open Source Program Office, marking a significant step toward democratizing AI-powered systems for developers and beyond.
However, as Jackie Brosamer, VP of Data & AI Platform Engineering at Block, told VentureBeat in a video call interview recently, the goal is to allow users to build agents that can cut across multiple software vendors, transporting user data, summarizing, and taking action on behalf of the user in multiple apps.
“I think the advantage of Goose is its ability to work across different systems. It’s not just limited to Google Drive—it can also integrate Google Drive with Slack, for example. One of our business team members finds it especially useful for figuring out what they did last week because they’re in 40 hours of meetings and can’t remember. And so it’s able to hook into any of these different systems, versus having to use one agent for Google, one agent for Slack, and kind of marry those together yourself.”
Google is now available under a permissive Apache 2.0 license, allowing it to be used for any conceivable project from the commercial to research.
It aims to unlock innovative workflows by enabling interoperability between user interfaces, AI models, and the systems they interact with.
Goose is designed to work with Anthropic’s open source Model Context Protocol (MCP)—a standardized suite of APIs that facilitate connections to tools like content repositories, business applications, and development environments.
Building the blocks of an open AI ecosystem
According to Dhanji Prasanna, chief technology officer (CTO) of Block, the open-source release is meant to inspire a wave of invention across industries. “Making goose open source creates a framework for new heights of invention and growth,” Prasanna said. He emphasized that the framework has already freed Block engineers from repetitive tasks, enabling more impactful work. “As more people build with AI, we will unlock new ideas and advantages that can collectively reduce toil and give people time back to be their most creative selves,” he added.
The software is modular by design, allowing developers to integrate Goose into their systems or build custom interfaces.
Goose’s open architecture also lets users select the LLM of their choice, offering flexibility for different use cases.
Bradley Axen, AI Tech Lead at Block, emphasized this adaptability, stating, “We’ve built Goose to work with any large language model, from OpenAI to open-source options. This adaptability allows us to combine strengths from different providers for optimal performance.”
Beyond engineering
Goose’s initial focus is on software engineering, where it functions as an agent capable of automating time-consuming tasks. Operating in real time within development environments, it can autonomously:
• Search and navigate codebases.
• Read, write, and edit files.
• Install dependencies and run tests.
• Refine outputs and handle additional tasks with minimal oversight.
This flexibility has the potential to save significant time for engineers while improving productivity. However, Block envisions broader applications, from streamlining creative processes like music composition to enhancing personalized e-commerce experiences.
Goose’s capabilities are enhanced through Block’s collaboration with Anthropic, a leading AI safety and research company. Together, they developed MCP, which enables Goose to interface with a growing ecosystem of tools and systems.
The open-source nature of Goose fosters community-driven innovation. As Brosamer explained, “Open-source development enables an ecosystem where new integrations can come from outside the company. Users don’t have to wait for us to release features; they can innovate on their own.”
Axen underscored the philosophy behind Goose’s release: “We’re not monetizing Goose directly—it’s fully Apache licensed. Instead, we aim to develop products that work seamlessly with agents, benefiting from the open-source ecosystem.”
As for privacy and security and considerations, the open source nature allows for flexibility of deploying Goose on premises or in virtual private clouds.
“We definitely do not have anything in the middle of Goose usage — no calls to our servers,” Axen said.
“You’re able to host it wherever,” Brosamer noted. “And as a financial company that’s super concerned about user data, that’s always been one of the reasons we’re really interested in open source models: rather than having to pass our data or users’ data over the internet, which comes with a lot of risk, we’re able to bring the models to where the data is already hosted.”
For more details, including installation and contribution guides, visit Goose’s Github website.
The post Jack Dorsey is back with Goose, a new, ultra simple open source AI agent building platform from his startup Block appeared first on Venture Beat.
