Sourcegraph unveils Cody 5.1, a free code AI tool that can write entire files and tests

Sourcegraph unveils Cody 5.1, a free code AI tool that can write entire files and tests

Join top executives in San Francisco July 11-12 to hear how leaders are integrating and optimizing AI investments for success. Learn more

source graph, a leader in universal code finding and AI-assisted software engineering, today announced the release of Cody version 5.1, a major update to its AI coding assistant. The new release gives Cody greater insight into code context through repositories and better automation capabilities, allowing him to generate code, fix bugs, and refactor projects with less human intervention.

In an exclusive interview with VentureBeat, Sourcegraph CEO Quinn Slack discussed the new Cody desktop app and its ability to create context for AI code. By allowing developers to point their local code to Cody, the app can better understand the code base and even write entire tests and files. “Cody now has a deep understanding of code bases that allows developers to trust him to write entire files, fix bugs, and answer questions about code they’ve never even seen,” he said.

The main improvements in Code 5.1, according to Slack, are the ability to understand context across multiple repositories in a code base, and new automation “recipes” that can perform more complex software engineering tasks such as performance tuning, code odor fixing and generating unit tests. Developers get online access to Cody via a chat interface in their code editors, and Cody can now make changes directly to the code.

Cody 5.1 poses challenges for competitors like GitHub’s Copilot, an autocomplete tool that relies primarily on a developer’s current code context. “Copilot was great when it released two years ago, but not much has changed,” Slack said. “Anyone who has used ChatGPT knows that AI could do a lot more than sophisticated autocomplete.”


Transform 2023

Join us in San Francisco July 11-12, where top executives will share how they integrated and optimized AI investments for success and avoided common pitfalls.

subscribe now

Improved autocomplete and new recipes

The broader code context comes from Sourcegraph’s strengths as a leading code research and analysis platform, which Cody now draws upon. “Cody benefits from 10 years of building a leading code search engine,” Slack said.

Multi-repository context and more advanced natural language understanding allow Cody to handle ambiguous questions and requests, as well as write idiomatic code while learning patterns across a code base.

Slack explained that the desktop application generates a local code graph by indexing the code for search and building embeds for semantic search, allowing the editor to communicate with the app for context when developers use Cody.

“Cody is the first code AI to autocomplete based on the context of the entire repository, using semantic search based on embeds,” Slack told VentureBeat. “This means that Cody can generate better code that uses more APIs and idiomatic usage patterns of your codebase, versus GitHub Copilot and others that only use recent files and open tabs.”

Go beyond autocomplete

Slack also said that Cody 5.1 goes beyond autocomplete and can do higher level coding tasks like writing entire files, tests, docstrings, variable names, release notes, pull request descriptions, performance tuning, patching of code smells and answering questions about the codebase.

“Cody can explain, write, fix, and refactor code using your code base’s APIs, documentation, and usage patterns,” Slack said. “This goes way beyond autocomplete or prompt engineering. It’s only possible because Cody provides the context of your code to a powerful LLM (Large Language Model), so that he can do higher-level coding tasks.

Cody 5.1 also introduces new features such as online chat, which allows developers to ask questions and request changes on specific regions of code files; support for JetBrains IDEs, such as IntelliJ, PyCharm, WebStorm; and the Cody desktop app, which makes it easy for people to use Cody on their own private code in their own editor and chat UI.

Cody 5.1 is free for developers on both public and private code, with a generous rate limit. Sourcegraph only charges for team/company/company features or for exceeding rate limit. Sourcegraph Enterprise Server users must upgrade to version 5.1 to get the new Cody features.

According to Slack, Cody 5.1 uses more context from the entire code base and multiple repositories, as well as a more powerful language model, Anthropic Claude, to generate more accurate and consistent code suggestions.

The future of artificial intelligence in programming

Discussing the role of the open development community in contributing to Cody 5.1, Slack said, “Cody is open source. It’s Apache 2.0 and we’ve had a lot of contributions. I think we have 20 different contributors and hundreds of people on our Discord so far.

He also stressed the importance of having an open platform and API for developers to get the most out of a product like Cody.

As for the future of AI in coding, Slack envisions a future where AI agents can take multiple steps to improve code without human intervention. However, he believes building trust between developers and AI is crucial before reaching that stage.

“We’re really excited about (the future of AI in programming),” Slack said. “We are monitoring this very closely. We’re also building it with Cody. Now, we have to tread carefully, because right now you have AI code that writes code where no human reviews it, so this is where the limits on adoption are broken.

Sourcegraph’s vision for AI

The new release is an important step for Sourcegraph in its vision of artificial intelligence that can automate complex, multi-step software engineering tasks. The company needs to tread carefully, Slack said, to ensure the AI ​​generates appropriate code and results for existing code bases in enterprise settings. But progress towards more advanced automation could significantly increase developer productivity.

“Our approach — more and better context, more powerful LLM — is different than other AI code autocomplete tools that optimize for limited context and small models,” he said. “We are optimistic that this ceiling approach will definitely outperform the minimum approach.”

Sourcegraph is a San Francisco-based company founded in 2013 by Slack and Beyang Liu. The company has raised $248 million in funding from investors including Sequoia Capital, Andreessen Horowitz, Insight Partners and Geodesic Capital. source graph annual revenue are estimated to be between $10 million and $50 million and has about 160 employees. Sourcegraph’s clients include Amazon, PayPal, Lyft, Uber, Yelp, Cloudflare, Plaid, GE and Atlassian.

VentureBeat’s mission it is to be a digital city square for technical decision makers to gain insights into transformative business technology and transactions. Discover our Briefings.