2025 is gonna be a year of agents. I'm more bullish on it than ever.

The first-ever Virtual Code AI Summit brought together some of the sharpest minds in software engineering, AI, and developer productivity and showcased how enterprises are adopting AI tools to manage growing code complexity, accelerate workflows, and fundamentally change the role of developers.

Across sessions led by experts from Booking.com, Stripe, Palo Alto Networks, Netflix, Leidos, and many others, a few overarching themes emerged:

1. AI is an essential assistant, not a replacement—empowering developers to focus on high-value work.
2. Automation with AI coding agents is already saving enterprises thousands of hours annually.
3. AI’s role is expanding across the entire software development lifecycle, from debugging to migrations to personalized tooling.

Read on to get a deeper look at these key insights and what they mean for the future of software development. Or, if you want to see all the recordings, click below to access the sessions.

VIEW SUMMIT RECORDINGS ->

AI is an essential developer assistant, not a replacement

One resounding theme throughout the summit was that AI tools and coding assistants are best used as force multipliers, not replacements for human developers. Leaders across companies emphasized how AI enables engineers to work faster, offloading repetitive or low-value tasks.

Bruno Passos, Group Product Manager at Booking.com, described the mindset shift needed for widespread adoption:

This isn’t about replacing developers. It’s a super assistant—someone over your shoulder, pointing out the good, the bad, and helping you move forward faster.

AI-powered tools like Sourcegraph’s Cody became integral to Booking’s development workflows, improving merge request speeds by up to 40% for power users. The impact wasn’t limited to junior engineers; senior developers also found that AI helped them focus on bigger, more complex (and interesting) challenges.

Similarly, Leidos, a leader in government contracting, shared how AI is uniquely suited to address the challenges of working with real government codebases, which often involve massive monoliths, decades of technical debt, and high-security constraints.

As Leidos’ Chief AI Officer Ron Keesing noted during his session:

AI helps us parse through codebases that would take humans weeks or months to untangle. With AI, discovery, navigation, and modernization become significantly faster and safer.

The takeaway is clear: teams that integrate AI into their workflows experience real productivity boosts, while developers who resist risk being left behind.

Building AI coding agents to automate repetitive work

One of the most exciting trends discussed at the summit was the rise of AI coding agents—automated tools designed to handle well-defined, repetitive tasks across the development lifecycle.

Quinn Slack, CEO of Sourcegraph, shared real-world examples of enterprises already seeing measurable ROI from coding agents:

• Test Automation: Companies use AI agents to generate unit tests across their codebases, significantly improving test coverage with minimal manual effort.
• Code Migrations: Booking.com is using agents to accelerate its shift from monoliths to microservices, reducing the manual burden of large-scale refactoring.
• Security Updates: ConService built an AI agent to automate dependency upgrades, saving over 1,000 hours annually after just 15 hours of setup.

Quinn Slack summarized the opportunity:

Think of AI coding agents like 5,000 smart interns. The key is to focus on small, well-scoped tasks—things that happen hundreds or thousands of times a day.

The success of these agents stems from their specificity—automating narrow, repetitive tasks rather than attempting to replace end-to-end workflows. Enterprises that embrace this mindset are unlocking massive time savings while improving code quality and consistency.

Expanding AI’s role across the software development lifecycle

While code AI adoption has largely focused on coding and code review, the summit revealed how AI is expanding its impact across the entire software development lifecycle (SDLC).

Adam Berry, now at Netflix, emphasized the untapped potential in areas like observability, debugging, and DevOps automation:

We’re just scratching the surface. AI tools that triage logs, reduce noise, and provide contextual insights will lift cognitive load and free teams to focus on higher-value work.

Gunjan Patel of Palo Alto Networks offered a vision of AI as a partner, helping teams tackle large-scale challenges—like replatforming legacy systems—and even enabling hyper-personalized developer tooling:

AI is democratizing software creation. Tools like Sourcegraph make it possible to build custom solutions that meet your exact needs, tailored to your workflows.

From CI/CD pipelines to security scanning to bug triage, AI is poised to deliver compounded productivity gains across every phase of the SDLC. The future isn’t just about coding faster—it’s about automating workflows, improving quality, and enabling developers to focus on creative, complex challenges.

The path forward: AI-augmented teams tackling bigger problems

The Virtual Code AI Summit made one thing clear: the future of software development belongs to AI-augmented teams. AI coding assistants and agents are eliminating toil, streamlining complex tasks, and giving developers the tools to solve problems at scale.

Quinn Slack captured this shift perfectly:

We’re automating the low-level tasks nobody likes doing, so developers can focus on what humans do best—solving big, meaningful problems.

From government agencies modernizing decades-old systems to tech leaders accelerating developer workflows, the potential of AI to reshape software development is limitless. The organizations that measure impact rigorously, build trust with AI tools, and embrace automation will lead the way.

If you want to access all these great talks and more, head to the summit page and access all the content on-demand.