Enhancing Developer Productivity with Duet AI: Turing @ Google

Google Next 2023 explored predictions and shared insights about the future of AI, cloud, and collaboration. Two of the event speakers, Vijay Krishnan, Founder and CTO at Turing, and Preston Holmes, Outbound Product Manager at Google Cloud, explained how advancements in AI and cloud would help companies optimize their operations by maximizing developer productivity.

Here are the key takeaways from the event.

Holmes began by touching upon the two generative AI-based products launched by Google: Duet AI and Vertex AI. “Duet AI is a place for you to consume the value of [Google] spending the time and effort building generative AI into [Google’s] products and services, whereas Vertex AI is a platform that enables you to be the producer of value, using generative AI in your own products and services,” he said. Here’s more on the two products.

Powered by Google’s latest generative AI foundation models, Duet AI is a powerful collaborator that can act as a coach, thought partner, source of inspiration, and productivity booster—all while ensuring every user and organization has control over their data.

With Duet AI, developers can use natural language to quickly understand complex code and APIs, follow coding best practices, quickly generate unit tests, and more. Duet AI’s code and chat assistance are available in development environments such as the Google Cloud console, Cloud Workstations, and Cloud Shell Editor. In addition, through Cloud Code IDE extensions, developers can use Duet AI in third-party IDEs such as VSCode and JetBrains IDEs like CLion, GoLand, IntelliJ, PyCharm, Rider, and WebStorm.

Vertex AI is a machine learning (ML) platform that lets developers train and deploy ML models and AI applications and customize large language models (LLMs) for use in AI-powered applications. Vertex AI combines data engineering, data science, and ML engineering workflows to enable software teams to collaborate using a common toolset and scale their applications using the benefits of Google Cloud.

Companies worldwide are undergoing rapid AI-driven transformation, with new AI-based products launching almost every day. “Products like Duet AI and Vertext AI are still being built on a moving train of innovation,” Holmes said.

These are emerging technologies, so naturally, very few providers have deep expertise on them. Therefore, your business needs a service provider that is adept at leveraging these new technologies to help you reach your goals.

Turing’s intelligent Talent Cloud can transform your business through AI.

But how exactly can Turing help you embrace cutting-edge AI-driven technologies? Let’s find out.

Addressing the question, Turing CTO Krishnan began with a short introduction about the company. “At Turing, we have helped over 1,000 companies and 100 large enterprises in two different ways. One: we help companies with fully managed software projects. Two: we help companies build remote distributed teams of pre-vetted developers.”

Krishnan further explained that Turing’s Talent Cloud allows rapid scaling of high-quality teams due to three factors. “First, we have a global pool of over 3 million software developers from 100+ countries in every possible time zone. Second, we thoroughly test developers on technical skills in various tech stacks, programming skills, soft skills, and leadership skills. Last but not least, we build very detailed developer profiles. These allow us to build good machine learning algorithms that find the right developer for any job. Needless to say, because we cast such a wide net, we cover every possible skill stack, industry, and domain,” he added.

He continued, “Turing always aspires to help software development teams to produce maximum value for clients. For this, we have been looking at the best tools in the market to help developers do their best.”

As a result, Turing started working with Google Cloud Platform (GCP) Workstation.

Here’s a breakdown of Turing’s journey with Duet AI for AI-accelerated delivery.

• Turing explored GCP Workstation for secure remote development and discovered a powerful code autocompletion tool—Duet AI.
• Turing integrated Duet AI and GCP Workstation for secure, high-performance coding across multiple IDEs.
• Developers loved how Duet AI's advanced autocompletion increased their productivity.
• Duet AI Chat was seamlessly incorporated via environment variables and was acclaimed by developers for enhancing daily workflows.

Turing’s experiment with Duet AI

Turing recently conducted an experiment to test how Duet AI influenced the productivity of its developers. The experiment included full-stack developers, data scientists, and data and ML engineers. The company measured pull request (PR) merges 4 weeks before and after Duet AI access. Here are the results:

• Turing saw a 33 percent productivity improvement in PR merges per developer for the 4-week time period.
• The company also saw early evidence that optimally training developers to use Duet AI leads to even higher productivity gains.

Turing’s matching platform is built on Vertex AI, making the company adept with GCP’s AI in its operations. In fact, with the help of Vertex AI, Turing has seen:

• Faster time-to-production: Deployment times were slashed by 80 percent, boosting efficiency and speed.
• Unified model management: The seamless creation, deployment, and management of models created a hassle-free experience.
• Code-free model training: Streamlined processes reduced model training time without the need for complex coding.
• Accessible tools: From novice to expert, Vertex AI offered tools for every skill level.
• Enhanced efficiency: Improved prediction efficiency led to better-informed decisions.
• Robust monitoring: Reliable monitoring ensured model security and flexibility.
• Empowering MLOps: Turing harnessed the true power of MLOps to connect enterprises with 3 million+ remote developers.

Krishnan further elaborated on Turing’s AI-driven software development flow. He mentioned that GPT helped design the [software development] system, but it required good prompt engineering to do so. During this process, Copilot and existing IDEs focused on coding and debugging, including comments, unit tests, and creating PRs.

Krishnan also added that it was challenging to do end-to-end tests as the complexity of software increased. During this time, an agent-based workflow could greatly improve the end-to-end velocity to ship software. He explained that this process was still programming, but at a higher level of abstraction.

Krishnan briefly touched on Turing’s future with GenAI. He mentioned that the company will try out newer features, like codebase indexing and debugging, to increase productivity. This will support Turing’s objectives for AI-accelerated delivery and improving developer search and recommendation capabilities for faster and higher-quality filling of job openings.

Today, technologies like Duet AI and Vertex AI cover only a portion of Google Cloud’s products and services, but in the future they will be the entire GCP experience. As mentioned by Holmes, these two technologies will follow companies and developers through the entire development process.

At this time, only a few providers can provide expertise on these technologies to help companies maximize developer productivity.

And Turing is among the very few providers operating in this space.

Turing has proven engineering excellence, expertise, and experience in this emerging AI-accelerated development space. If you’re looking to reap the gains from Google’s GenAI tools by becoming an early adopter, Turing is here to help you.

Make your AI implementation goals a reality with Turing.

Your modern software engineering challenges deserve modern development methodologies. Turing Services will provide you with the right mix of AI, cloud, and application engineering to take you from legacy to industry leader.

Visit the Turing Services page to learn more.