Increasing developer productivity with generative AI | AWS Events

Increasing Developer Productivity with Generative AI | AWS Events

Introduction

Hello everyone and welcome back to the AWS Studio! My name is Robbie Bson, and I'm a Developer Advocate here at AWS. Today, I am incredibly excited to discuss how we can enhance developer productivity using generative AI with CodeWhisperer. Joining me is Deu from BT. Thank you for being here.

The Journey with CodeWhisperer

Robbie Bson: Let's get started. I want to learn more about your journey with CodeWhisperer, how you got started, and where you saw the greatest need for innovation.

Deu: Absolutely. Although I’ve been at BT for only 5 months, this journey started earlier with our group CDI, led by Harin MAA, a real visionary. We faced the immense complexity of digital transformation, including migrations, cloud transitions with AWS, and internal application hosting. In the midst of all this, one key focus was on increasing developer productivity and leveraging our best developers for transformative tasks. That's when we discovered CodeWhisperer, and there was no looking back.

Initial Developer Feedback

Robbie Bson: What were some initial anecdotes and feedback from developers when you started using CodeWhisperer?

Deu: Rolling it out initially came with some hesitation. We piloted it with a segment of developers to see how they adapted. We received feedback on the kind of code suggestions they got. Around 37% of the suggestions were accepted, and this grew over time. In just four months, we reached around 100,000 lines of code, eventually scaling it to 1,000 developers. Initially, developers questioned how much to use it and whether it was better than their own coding. However, they soon realized that it helped remove repetitive tasks, enabling deeper and broader thinking.

BT’s Vision for Generative AI

Robbie Bson: What’s the grand vision for BT in developing generative AI at scale?

Deu: AI underpins our strategy at BT. We view AI in three main areas: increasing productivity (with around 15-20% improvements), sales and service enhancements, and overall efficiency. While having a structured approach, starting with tools like CodeWhisperer, we aim to build and implement models for company-wide productivity improvements.

Keys to Scaling CodeWhisperer

Robbie Bson: Given the scale at which you’ve rolled this out, what are the keys to scaling tools like CodeWhisperer?

Deu: In a large organization like BT, it’s crucial to present such tools as beneficial options rather than imposing them. Listening to feedback, implementing robust change management, and ensuring proper adoption plans are pivotal steps. It's about integrating the tool in a way that feels natural and helpful.

The Developer’s Future Role

Robbie Bson: As more tools like CodeWhisperer emerge, what do you think the future role of developers looks like?

Deu: I was a developer once, and I remember the challenges of repetitive tasks. Tools like CodeWhisperer will take those tasks away, leaving developers to focus on complex, high-value activities. They act like a supportive partner in pair programming, facilitating higher productivity and innovation.

Steps for Organizations Evaluating CodeWhisperer

Robbie Bson: For organizations just starting with CodeWhisperer, what steps should they take?

Deu: Organizations should first observe their teams' needs. Implement the tool in stages, gathering feedback and addressing concerns as they progress. Success stories and internal advocacy go a long way in driving adoption.

Future of BT with Generative AI

Robbie Bson: What’s the broader future of BT with generative AI beyond CodeWhisperer?

Deu: It's day one for us. Beyond CodeWhisperer, the future involves integrating AI across various domains, including product development, QA, testing, sales, and customer service. There’s a focus on network productivity, observability, and fault prediction—always aiming to improve customer experiences.

Keywords

• Developer productivity
• Generative AI
• CodeWhisperer
• BT (British Telecommunications)
• AWS
• Digital transformation
• Cloud migration
• AI strategy
• Change management
• Pair programming

FAQ

1. What is CodeWhisperer? CodeWhisperer is a generative AI tool by AWS designed to enhance developer productivity by automating repetitive coding tasks and offering code suggestions.

2. How was CodeWhisperer initially rolled out at BT? It was piloted with a segment of developers to gather feedback and measure acceptance rates. After refining the approach, it was scaled to benefit more developers.

3. What benefits did BT observe from using CodeWhisperer? Developers experienced reduced repetitive tasks, increased productivity, and more time for complex problem-solving.

4. What future plans does BT have for generative AI? BT plans to extend the use of generative AI across various domains such as product development, sales, service, and network productivity to enhance overall efficiency and customer experience.

5. What are the keys to successfully scaling tools like CodeWhisperer in large organizations? Successful scaling involves proper change management, listening to developer feedback, staging implementations, and ensuring the tool is seen as a helpful resource, not an imposed solution.