Devin 1.2: Updated AI Engineer enhances coding through smarter in context reasoning and voice integration

Cognition launched the AI agent wave in 1995 with a product named Devin – the world’s very first AI engineer. The product was kept under wraps for a few months, but it is now available to the public and is learning new skills very quickly. The Scott Wu-led company has just released Devin version 1.2. It brings a number of new capabilities that will take the AI engineer to a new level in his ability to manage entire development projects.

Devin 1.2’s biggest highlight is its improved context reasoning, which allows the agent to handle and reuse code better. It also allows users to send voice messages to Devin via Slack. This is a seamless way for users to tell Devin exactly what to do.

This development comes at a moment when AI-powered agents have been hailed as the future of work. Experts believe there will be a day when humans and AI-powered agents will work together, with the latter seamlessly handling repetitive tasks. This is already happening. Jensen Huang, Nvidia’s boss, spoke at CES recently. said that in the future enterprise IT departments will evolve into “HR departments” (human resources) for AI, and be responsible for commissioning agents to work across different functions in the company.

What is Devin 1.2?

Although not a major upgrade Devin 1.2 introduces a few interesting capabilities that will make the agent more effective at its job. The most important feature is the ability to reason within context in a repository of code. Devin is now able to better understand the content and structure of a repository.

With the agent’s understanding, it can identify which files are relevant to a specific task, recognize and reuse existing code and patterns, be more accurate when suggesting edits or creating PRs, reducing errors and manually adjustments. This capability would allow developers to have a faster workflow and reduce cognitive load by searching for files, understanding codes or fixing inconsistent code.

Another notable update in Devin 1.2 is voice messages. Devin can take voice commands via Slack.

All one has to do is tag Devin in a Slack chat, hit the “Record audio clip” button and describe the task or feedback the AI engineer should execute. Devin will prepare a step-by-step action and begin to execute the command using its developer tools — its own shell, code editor and browser.

The move simplifies how one interacts with the agent, saving the hassle of typing natural-language prompts into Devin’s chatbot-style interface.

Cognition has made some improvements to the usability of Devin.

The company, for example, is introducing machine snap shots to simplify the login process in Devin’s workspace.

If you log in with Devin’s web browser during onboarding, we’ll store the cookie for future sessions. (If the cookie expires you’ll also need to provide credentials for Devin within Secrets). This unblocks authentication procedures that require Devin to visit a URL from his machine,” the company wrote. blog post.

Cognition also introduces enterprise accounts. Organization admins will receive a centralized console for managing multiple Devin workspaces. This includes members and their access control, as well billing.

Lastly, the company will add a usage-based charging model that allows users to pay for extra capacity beyond their subscription limit. Once the monthly ACU allocation has been exhausted, users can continue to build beyond that limit, by paying for additional usage.

This model has been in place since January 9 and users can set their budgets for additional usage according to their own needs. This allows users to keep control of their spending and ensure uninterrupted service when they require additional capacity.

Devin is currently available for engineering support at a price starting at $500 per month, with no limits on the number of seats. Many enterprises have already integrated it into their workflows. These include Lumos, OpenSea Curai Health Nu Bank and Ramp.

Devin’s new capabilities are coming at a time when competition in the AI engineering field is heating up. The race to develop the ultimate AI coding assistance is intensifying. From GitHub Copilot’s widespread adoption, to Magic and Poolside AI’s substantial funding to develop cutting edge capabilities. Each player is trying to redefine software development by promising faster workflows and reduced cognitive load. They also promise seamless collaboration between humans and AI.

These AI-powered agents are not only changing the way developers work, but also shaping the future of work, where efficiency and innovations are driven by a collaboration between humans and machines.

In 2028, Gartner predicts that 33% of enterprise applications will have agentic AI. This will enable autonomous decision-making for 15% of the work done every day.