Jack Dorsey is back with Goose, a new, ultra simple open source AI agent building platform from his startup Block

The identity of Twitter on the social network on the theme of birds may have been X-ED by the new owner Elon Musk, but that did not stop one of his co-founders, Jack Dorsey, to take New names of birds for a new project.

The other company in Dorsey BlockThe parent of Point of Sale Service Square, Mobile Payments System Cash App, Music Streaming Service Tidal and other financial tools focused on technology, Today launched the goosea free, Open Source Framework who seeks to simplify the process of building an AI agent (or many agents) with almost all models of language (LLM) conceivable as intelligence on the backend, whether in depth or a model Own of Openai, Google, whether it is and anthropogenic.

There is less than 24 hours, Dorsey has launched the launch of the goose in force by Publication on X: “Any open source.”

The agents that Goose is designed to build are mainly focused on the development of software, and in a blog post announcing the new framework, the Advocate Bloc developer Adewale Abati wrote that the frame has already helped the following tasks:

• Perform code migrations such as EMPER to react, Ruby to Kotlin, Prefect-1 in Prefect-2 etc.
• Immerse yourself in a new project in an unknown coding language
• Transition An injection injection code base to injection based on the manufacturer in an addiction injection framework.
• Perform performance references for a build order using a build automation tool
• Increase in code coverage above a specific threshold
• API scaffolding for data retention
• Creation of data companies monitors
• Deletion or addition of functionality flags, etc.
• Generate unit tests for a functionality

The announcement was made today by Open Source Block Program OfficeMarking an important step towards democratizing it of systems fueled by AI for developers and beyond.

However, as Jackie Brosamer, Vice-President of data and engineering of the AI ​​platform at Block, recently told Venturebeat in a video call interview, the objective is to allow users to create agents Who can reduce several software suppliers, transport user data, summarize and take action on behalf of the user in several applications.

“I think the advantage of the goose is its ability to work on different systems. It is not limited to Google Drive – it can also integrate Google Drive with Slack, for example. One of the members of our sales team finds it particularly useful to determine what they did last week because they have 40 hours of meetings and do not remember it. And so he is able to hang on to one of these different systems, compared to using an agent for Google, an agent for Slack and in a way marry them yourself. “”

Google is now available under a Apache 2.0 licenseallowing him to be used for any imaginable project of research advertising.

It aims to unlock innovative workflows by allowing interoperability between user interfaces, AI models and the systems with which they interact.

Goose is designed to operate with the context protocol of the Open Source of Anthropic (MCP) – A standardized API series that facilitates connections to tools such as content benchmarks, commercial applications and development environments.

Build the blocks of an open AI ecosystem

According to Dhanji Prasanna, Director of Technology (CTO) of Block, the open-source liberation aims to inspire a wave of invention in industries. “The manufacture of the open source of the goose creates a framework for new heights of invention and growth,” said Prasanna. He stressed that the framework has already released engineers from repetitive tasks, allowing more impactful work. “While more and more people are building with AI, we will unlock new ideas and advantages that can collectively reduce work and give people time to be their most creative self,” he added.

The software is modular by design, allowing developers to integrate the goose into their systems or to create personalized interfaces.

The Goose Open architecture also allows users to select the LLM of their choice, offering flexibility for different use cases.

Bradley Axen, head of AI Tech in Block, underlined this adaptability, declaring: “We have built Goose to work with any large, from Openai model to open-source options. This adaptability allows us to combine the forces of different suppliers for optimal performance. »»

Beyond engineering

Goose’s initial accent is on software engineering, where it works as an agent capable of automating long tasks. Operating in real time in development environments, it can be autonomous:

• Search and navigate in code bases.

• Read, write and change files.

• Install the outbuildings and run the tests.

• Refine outings and manage additional tasks with minimum monitoring.

This flexibility has the potential to save a lot of time for engineers while improving productivity. However, Block envisages wider applications, rationalization of creative processes such as musical composition to improve personalized experiences of electronic commerce.

Goose’s capabilities are improved thanks to blocking block with Anthropic, a main IA security and research company. Together, they have developed MCP, which allows the goose to interface with an increasing ecosystem of tools and systems.

The open source nature of the goose promotes innovation focused on the community. As Brosamer explained, “open-source development allows an ecosystem where new integrations can come from the outside of the company. Users do not have to wait for us to publish features; They can innovate by themselves. »»

Axen underlined the philosophy behind the exit of Goose: “We do not go directly to the goose – it is entirely authorized by Apache. Instead, we aim to develop products that work perfectly with agents, benefiting from the open-source ecosystem. »»

With regard to confidentiality, security and considerations, open source nature allows the flexibility of the deployment of goose on premises or in virtual private clouds.

“We certainly have nothing in the middle of the use of the goose – no call to our servers,” said Axen.

“You can host it everywhere,” said Brosamer. “And as a financial company which is very concerned about user data, this has always been one of the reasons why we are really interested in open source models: rather than having to transmit our data or our users On the Internet, which comes with a lot of risk, we are able to bring the models where the data is already hosted. »»

For more details, including installation and contribution guides, visit Goose’s GitHub website.