‘Developers are moving from writing code to orchestrating AI agents’: AWS’s Jeff Barr

Cloud computing major Amazon Web Services (AWS) recently held the 10,000 AIdeas Competition that required builders from around the world to pitch an AI app idea and build it from scratch. “From day one, our mission was simple. To make powerful technology accessible to every developer,” said Jeff Barr, vice president and chief evangelist, AWS.

“We envisioned supporting everyone, from a developer working in a dorm room, a garage, or a coworking space, to startups and large global enterprises. We wanted developers everywhere to have access to the tools and infrastructure needed to transform great ideas into reality and scale them globally,” Barr told indianexpress.com.

He said the 10,000 Aldeas competition is a natural extension of AWS’s core mission. A key requirement of the competition was that developers use Kiro, AWS’s agentic development environment, to build AI applications that solve real-world problems. “We also challenged them to do so within the AWS Free Tier, which allows developers to start building without any upfront cost commitment.”

The competition was held from December 2025 to April 2026 and attracted thousands of aspiring developers from over 115 countries.

‘Exceptional quality’

Consequently, Barr shared that the young developers exceeded expectations and impressed all with the breadth of their imaginations, the scale of their ambitions, and the quality of what they created. These developers created apps that address challenges across domains spanning healthcare, sustainability, education, and everyday life. “Beyond creativity, what stood out was the exceptional quality of the applications.”

According to Barr, the participants weren’t merely creating prototypes; they were building production-ready applications with Kiro’s spec-driven development approach. “They weren’t just experimenting with prompts; they were developing structured, reliable, and deployable software.”

The gamut of applications being developed at the competition, according to Barr, indicates a broader shift across the developer community. They are moving farther from writing software line by line to working at a higher level.

Story continues below this ad

Barr believes this also allows developers to focus on understanding users and solving domain-specific problems rather than spending time on syntax and implementation details.

For healthcare and scientific accuracy

Computer engineering students Shambhvi Sharma and Om Singh, the global winners of the 10,000 Aldeas Competition, also joined the conversation to discuss their apps.

Sharma and Singh, along with their teams, build AI-powered applications for Parkinson’s detection and for scientific claim verification, respectively. Sharma, along with Yash Aggarwal, created a multimodal Parkinson’s screening platform called NeuroVoice that tracks patterns over time to detect subtle neurological changes early.

“NeuroVoice is an AI-powered neurological monitoring platform. A patient sends a simple voice note through WhatsApp or our native app. The dataset we trained on extracts 26 acoustic features per voice sample, runs facial analysis, motor assessments, sleep and medication tracking, and builds a longitudinal biomarker picture over time,” Sharma said, adding that the app does not diagnose but flags trends that could aid in clinical follow-ups.

Story continues below this ad

When asked about the inspiration behind the app, Sharma said she came across research papers on acoustic and facial biomarkers in neurological diseases and found that the condition left measurable markers in a person’s voice and facial movements, often years before a formal diagnosis is possible. “The science was there. The gap was that nobody had made it accessible,” said Sharma.

On the other hand, Om Singh, along with Utsav Singh and Vikas Tiwari, built AEST, a claim verification platform that validates scientific claims against over 1.2 million peer-reviewed papers across 8 domains. “ASET is an evidence verification engine designed to help people discover the evidence behind the information they consume. We were not aiming to build something very big but rather solve a small problem that everyone is facing today,” said Om.

Om said information is generated everywhere in the form of articles, research papers, videos, social media, and AI chatbots. Responses by AI do not always ask the user to trust specific sources; instead, they generate an answer.

He added that ASET searches a corpus of scientifically validated data, breaks claims into smaller parts, and provides a detailed analysis of what the text implies. “ASET is available in multiple forms. It is not a platform. It is an engine that can be used in many ways.”

Story continues below this ad

The winning mix

Considering hundreds of thousands of entries, when asked what mechanism was in place to judge the best of the lot, Barr said the competition had a judging team that spent considerable time reviewing each entry. “Every participant prepared their own content, typically including a blog post and a video.”

Barr noted that one of the most remarkable things about the event was the breadth of skills demonstrated by the participants. The young developers had to identify an opportunity, imagine a compelling idea, and work with Kiro to turn that idea into a specification. The AWS executive feels another skill was equally important. “They needed strong communication and marketing abilities, not just to build something amazing, but to explain it clearly and effectively to the world.”

Barr also gave his nuanced take on the role of full-stack developers who work across all layers of technology. The chief evangelist said there is another dimension to that concept, as, according to him, the most successful participants were ‘broad problem-solvers who also possessed communication and marketing skills’.

India is home to one of the largest and fastest-growing developer communities. When asked what role he sees Indian developers playing in the next phase of AI innovation, Barr shared that they stand out for their combination of deep technical talent, a strong culture of experimentation, and a focus on solving real-world problems at scale.

Story continues below this ad

“The AIdeas competition made clear that India’s developer community isn’t just technically strong but also deeply engaged with some of the most pressing issues today across industries.”

AI-generated code and scepticism

With the likes of OpenAI, Anthropic, and Google pumping billions of dollars into developing advanced AI tools capable of coding, organisations are increasingly deploying these tools to automate workflows. When asked whether developers are becoming more comfortable trusting AI-generated code or whether scepticism persists, Barr said that developer confidence is increasing as the technology continues to improve.

“Less than three years ago, you’d prompt these tools and often get code that didn’t compile. Developers reviewed every line, corrected the gaps, and treated the output as a rough first pass. However, there has been a significant shift in that approach,” Barr said. “Developers are moving from writing every line themselves to orchestrating intelligent agents that can reason, plan, and execute.”

Barr said that, for him, the best way to get past hesitation is real, hands-on experience. The executive said that he had a validator tool on his task list for over a year but never got around. But that changed when he used Kiro, which handled everything independently with just a simple prompt. “It parsed the document, identified each section, determined appropriate rules, and produced a clean programme. It’s exactly the kind of moment, where a genuine task is resolved with minimal effort, that converts sceptics into advocates.”

Story continues below this ad

Barr believes that as AI tools become more capable, developers will spend less time on implementation details and more time on architecture, problem-solving, and delivering business outcomes.