How Augment Code is redefining developer velocity with AI

What developers actually want from AI

When Guy and his team began exploring how AI could boost productivity, they noticed something surprising: engineers weren’t asking for more code—they were asking for more understanding.

“When we started Augment, we realized developers don’t need another autocomplete plugin. They need something that understands their codebase and can reason about it intelligently.”

— Guy Gur-Ari, Co-Founder, Augment Code

Traditional coding tools predict the next line, but rarely grasp how everything connects. True velocity, Guy says, comes from reducing context switching.

“Our model understands structure—it knows how files connect, what calls what. When it changes something, it’s aware of the consequences across the system.”

Augment’s context engine helps AI agents navigate multi-million-line codebases across teams of hundreds or thousands—managing dependencies and architecture in real time.

That level of comprehension turns AI from a helper into a collaborator—one that understands not just what’s written, but how everything fits together.

Every company, Guy notes, has unwritten rules—naming patterns, file structures, problem-solving habits. Capturing that “vibe” is what separates a true AI teammate from a generic tool. Context isn’t just technical; it’s cultural.

Beyond assistance: How agentic systems redefine developer velocity

This evolution from code prediction to intelligent collaboration marks the rise of agentic systems—AI that doesn’t just complete tasks, but understands purpose and scope.

“This isn’t theoretical anymore. We already have customers where agents open and close tickets, debug, analyze logs, even respond to incidents,” Guy says.

These agents now manage entire workflows—accelerating development, testing, and incident response while reducing cognitive load for human engineers.

As AI generates and tests code faster than humans can review it, enterprises face a new limit: the code review bottleneck. Augment’s agents help close that gap by validating, debugging, and refining work in context.

Velocity is no longer bound by human attention—it’s multiplied by machine understanding.

The human shift: From coders to conductors

Even with these advances, Guy notes that adoption isn’t just a tooling challenge—it’s a behavioral one.

“Developers love typing—it’s tactile. It’s muscle memory. Asking them to stop and instead guide an agent—it’s unnatural, and it takes a while to adjust.”

Moving from writing code to orchestrating intelligent systems asks developers to think differently—less about syntax, more about intent and communication.

“Prompting is communication. The clearer you are, the better the model performs. It’s like leading a team—if you’re vague, you’ll get vague results.”

Tomorrow’s best engineers won’t just write code; they’ll lead collaborative systems. Developer velocity becomes less about output and more about clarity, orchestration, and shared context.

Building trust in autonomy

As AI takes on more responsibility, trust becomes the true measure of readiness. Transparency, oversight, and clear permissions ensure that autonomy builds confidence—not risk.

“When the agent gets access to more things—codebases, production, data—the risks rise. You need permissions, controls, guardrails. Without that, you’re taking big risks.”

For Guy, governance isn’t bureaucracy—it’s infrastructure. The more autonomy an agent has, the stronger the foundation of visibility and accountability must be.

Trust doesn’t slow innovation; it sustains it.

The next horizon: Customizable software and the agent experience

Looking ahead, Guy believes the falling cost of software creation will reshape how we build entirely.

“As the cost of code generation goes to zero, do we still live in a world where you need one company to build an app for millions of users to use, or can it become a lot more customizable? Maybe I can now create my own apps, or maybe take some sort of template of an app, but do a lot of customization on it to suit exactly my needs.”

If generating software becomes effortless, the challenge shifts from building for the masses to enabling individual customization.

“Right now there are all these walls between apps—I can’t easily connect my emails to my calendar or my logs in production to the code. They’re all living in silos. A term I really like that’s starting to come up is ‘AX,’ or agent experience—building your product not just for humans, but to be really easy for agents to use.”

This idea of agent experience reimagines software as a connected ecosystem—where humans and agents collaborate seamlessly. It’s a natural extension of Augment’s mission: making software not just faster, but smarter and more adaptive.

The leadership lesson: Adoption takes work

Even with all this progress, Guy reminds us that transformation isn’t just technical—it’s human.

“It’s rarely the tech that fails. It’s the change management. Tools work—the hard part is helping people change their habits.”

Organizations that succeed will treat adoption as a cultural evolution, not a rollout. Helping teams work confidently with AI means leading with empathy and patience.

Because real innovation happens when technology and people grow together.

Listen to the full conversation with Guy Gur-Ari, Co-Founder of Augment Code, on MindMakers to hear how contextual agents are transforming how enterprises build, test, and ship software.

Ready to explore Sendbird’s agent-ready infrastructure? Our background in communications APIs means we can break down the silos between all of your systems, so that our AI agents have all the data and context they need to support your customers. Contact sales to learn more.