Jack Dormy has returned with Goose, which is a simple new AI Opeen Agent Agent from his startup block

Bird-Twitter may have been directed by the new owner Elon Musk, but this did not stop one of its founders, Jack Dormy, who took the names of the new birds for a new project.

Other Dorsi company roadblockSquare Square Points-FF-SALE, applying mobile devices payment system, music flow service, and other technology-based financial tools, Today I launched a gooseFree, An open source frame Which seeks to simplify the process of building an artificial intelligence agent (or many agents) with any large language model (LLM) as an intelligence on the back interface, whether it is Dibsic Or a monarchy model such as Openai, Google and Anthropic.

Less than 24 hours ago, Dorsi disturbed the launch of a goat in reality Publishing on x: “Open source everything.”

The agents whose goat is mainly designed to develop software, and in a blog post announces the new framework, Block AdEWale Abati has been written that the frame has already helped in the following tasks:

• Code deportation such as EMber to respond, Ruby to Kotlin, Governor -1 to Governor 2 etc.
• Dive into a new project in an unfamiliar coding language
• The transfer of a symbol from the field -based injection to the creator -based injection within the framework of dependency injection.
• Performance standards for construction order using a construction automation tool
• Increase the code coverage over a specific threshold
• Souk the interface programming interface to keep the data
• Create Datadog screens
• Remove or add feature flags, etc.
• Generating unit tests for a feature

It has been announced today before Block Open Source Office officeOn the occasion of a big step towards the democratic character, the artificial intelligence systems of developers and beyond.

However, Jackie Brosamer, Vice President of Data & Ai Platform Engineering in Block, Venturebeat in a recent video call interview, is the goal is to allow users to build agents that they can cut through multiple software sellers and transmit user data, summarize, take and take them on behalf of the user in applications Multiple.

“I think the GooSe feature is its ability to work through different systems. Not only does Google Drive – Google Drive can also be integrated with Slack, for example. One of our team members finds that it is especially useful to know what they did last week because they are in 40 hours of meetings and they can not remember it.

Google is now under abuse Apache 2.0 licenseAllow it for any project that can be visible from commercial to search.

It aims to cancel the innovative workflow lock by enabling the interface between the user’s facades, artificial intelligence models and the systems that interact with it.

Goose is designed to work with Protocol context of the open model of anthropology (MCP)– A unified set of applications programming facades that facilitate communications with tools such as content warehouses, business applications and development environments.

Building the AI ​​Open Environmental System blocks

According to Dhanji Prasanna, Senior Technology Official (CTO) in Block, the open source version aims to inspire a wave of invention through industries. Brasana said: “Making Goose Open Source creates a new height of invention and growth,” said Brasana. He stressed that the framework has already released engineers from the tasks of repeated tasks, allowing more influential work. He added: “With more people building with artificial intelligence, we will open new ideas and advantages that can reduce toil collectively and give people time to return until they are more creative.”

The program is standard according to the design, which allows developers to integrate a goose into their systems or build dedicated facades.

GooSe’s open architecture also provides the LLM they choose, providing flexibility for different cases of use.

Bradeley Axen, AI Tech Lead in Block, stressed this ability to adapt, saying: “We have built a goose to work with any large language model, from Openai to open source options. This ability to adapt to combines the strengths of different service providers. For optimal performance.

Beyond engineering

Goose’s initial focus is on software engineering, as an agent who is able to automate the tasks consuming time. Work in an actual time in development environments, it can be independent:

• Research and navigating the symbol rules.

• Read files, writing and editing.

• Fix dependence and operate the tests.

• Refine outputs and deal with additional tasks with minimal control.

This flexibility has the ability to provide a great time for engineers while improving productivity. However, Block imagines wider applications, of simplifying creative processes such as composing music to enhancing personal e -commerce experiences.

Goose’s capabilities are enhanced by Block’s cooperation with humans, a leader in safety and research from artificial intelligence. Together, they developed MCP, which enables Goose from the interface with an increased ecosystem of tools and systems.

The open nature of the source of the Aws enhances the innovation that society drives. “Open source development allows an environmental system where new integration operations can come from outside the company. Users should not wait until we export features; they can innovate on their own.”

“We do not call directly a goose – it’s completely licensed from Apache. Instead, we aim to develop smoothly working products with agents, and benefit from the open source ecosystem,” Axen stressed the philosophy behind the edition of Oza.

As for privacy, security and considerations, the open source nature allows the flexibility of deploying goats in buildings or in virtual private clouds.

“Certainly we have nothing in the middle of the use of goats – there are no calls to our servers,” Axen said.

“You are able to host it anywhere,” Brosamer pointed out. “As a financial company, you feel very anxious about user data, this has always been one of the reasons that make us really interested in open source models: instead of having to pass our data or user data online, which comes with a lot of risk, we are able to bring models to where they are Hosting data already.

For more details, including installation and contribution evidence, please visit Oze Gybpip website.