Chasing AI-Native Software Development

I made a career decision that deepened and expanded my career journey and perspective. I joined Qodo as their Developer Relations Lead and it has been an incredible journey being fully immersed in AI for software development. And I discovered was something incredibly urgent: enterprise development teams are drowning in AI noise, hesitant about adopting AI tools that promises speed and ease of use, and running into bottlenecks at different checkpoints in the development lifecycle.

Most developers are struggling to use AI coding tools effectively.

The Uncomfortable Reality Check

Walk through any tech conference, scroll through Twitter/X, and you'll see the same pattern everywhere. Developers copy-pasting from LLM outputs or watching agents pump out code into their projects, crossing their fingers, and attempting to ship code. Confidence is low and vibes are immaculate (initially). The result, however, is AI slop, technical debt, and quite a bit of stress.

This is what the industry calls "vibe coding": letting AI run wild in your codebase, making development decisions based on AI’s capacity to churn out code rather than careful implementation. It's AI-assisted development without minimal intelligence or specification. When an approach to AI code generation is essentially "this feels right," it’s essentially assembling a house of cards.

When AI-generated code lacks proper context and guardrails, you get both the bugs and systemic security risks that can compromise entire orgs. This is a practice problem.

However… I'm seeing the industry shift from vibe coding to what I call "viable coding" out of sheer necessity. Teams are realizing that sustainable AI development requires structured frameworks, contextual understanding, and measurable outcomes. We're standing at a major inflection point in software engineering.

Goodbye AI Slop; Introducing Viable Coding

At Qodo, the R&D and product teams are architecting an agentic AI coding platform that understands your entire codebase, your team's conventions, and your organization's specific requirements.

This is the foundation needed to practice viable coding. AI development that's systematic, context-aware, and built for long-term maintainability. It's the antithesis of vibe coding, where every decision is grounded in data, patterns, and organizational knowledge.

Think less "AI as a tool" and more "AI as a principal engineer pair programming with you."

The transition from vibe to viable coding truly is about sustainable efficacy. Teams are discovering that what applies to fun side project development cannot be applied to real-world enterprise software development where your professional reputation (and the company’s reputation) is on the line. Viable coding compounds its benefits over time because it's built on principled foundations.

What Viable Coding Actually Means in Practice

Instead of asking AI to generate features and hoping it works, viable coding involves workflows where AI can:

Analyze architectural impact across multiple repositories before you write a single line of code

Generate comprehensive test suites that understand your existing patterns and edge cases

Perform deep code reviews that catch issues human reviewers consistently miss

Orchestrate entire deployment workflows while maintaining your compliance standards

This is what we're shipping at Qodo right now. And it’s exciting because I truly align with the vision. As a software developer myself, I will never stray far away from the need to have structure and pragmatism. I like realistic gains, not AI hype or fluff.

Solving the Enterprise Context Crisis

One of biggest challenge I've observed with using AI code assistants is context engineering. Most AI tools operate in complete isolation. They don't understand your codebase structure, your team's coding standards, or the intricate dependencies that define enterprise software architectures.

There are many other issues I see, but I’ll get into those in upcoming posts. To give a hint, they involve rethinking where agents should be utilized in the software development lifecycle.

The Framework I’m Testing

Through my brand new Agent Academy live event series, I'll been teaching attendees how to transition from reactive AI usage (fixing problems after they happen) to proactive AI orchestration (preventing problems before they exist) at scale.

The framework follows a clear pyramid structure:

Foundation Layer: Architecture

• Orchestrators that manage complex, multi-step workflows

• Specialized agents for different phases of the development lifecycle

• Shared memory systems that maintain context across interactions

• Compliance guardrails that enforce organizational standards

• Observability layers for auditing and continuous optimization

Automation Layer: Workflows

• Streamlined code review processes that catch issues early

• Accelerated test generation with real coverage

• Proactive regression detection before deployment

• Automated documentation that stays current

Experience Layer: Developer Impact

• Reduced cognitive load from repetitive tasks

• Enhanced focus on architectural and design decisions

• Faster onboarding for new team members

• Measurably improved code quality and security posture

Security Also Can't Be an Afterthought

As someone who's witnessed the aftermath of AI-induced vulnerabilities firsthand, I'm convinced that security is fundamental to the entire equation.

This is why I’m a firm believer that security-first design principles should be implemented from day one in AI coding:

Context-aware vulnerability detection that understands your specific threat model

Compliance validation at every step of the development process

Complete audit trails for all AI-generated changes

Native integration with existing security toolchains and processes

The Vision I'm Building Toward

My mission in this space goes beyond advocating for better tools. I'm working to establish the foundational practices that will define how enterprise engineering teams adopt AI responsibly over the next couple of years.

This includes:

Educational Leadership: Through conferences, technical content, and hands-on workshops, helping teams understand not just what's possible, but what's practical, secure, and sustainable.

Community Building: Creating spaces where developers can share real-world experiences, learn from actual failures, and collectively establish best practices.

Technical Evangelism: Working directly with enterprise teams to implement AI-native workflows that deliver measurable business value.

Industry Standards: Contributing to the broader conversation about responsible AI adoption in software development. (I’m hoping to get involved with CNCF AI/ML working groups in 2026!)

The Competitive AI Advantage

We're moving toward a world where the distinction between human-written and AI-generated code becomes completely irrelevant. What will matter is whether the code is correct, secure, maintainable, and aligned with business objectives. And not only that, but a world where developer experience with AI tools can impact engineering orgs’ ability to ship code faster with pristine quality and confidence.

Looking Forward

I foresee developers leveraging AI coding tools in a way that makes them better engineers, better collaborators, and better stewards of the technology. And that future is unfolding one workshop, meetup, research paper, blog post, conference talk, YouTube video, book, and customer-focused technical implementation at a time.

Feel free to connect with me and follow my work through technical and holistic deep-dives on LinkedIn, Twitter/X, and at conferences.