In a recent LinkedIn post, Rahul Kumar shifts the focus from traditional code reviews to a more fundamental challenge: the specification process. Kumar argues that many teams struggling with code quality and efficiency issues are, in fact, facing a “spec problem” rather than a code review problem. He highlights that the most costly disagreements often arise long before code is written, stemming from unclear requirements and misinterpretations of the project’s scope.
“Most teams don’t have a code review problem. They have a spec problem.”
The Root of Development Disagreements
Kumar elaborates on the common points of friction that occur by the time a piece of work reaches the pull request stage. These include fundamental questions about what is being built, whether requirements have been understood correctly, how new features integrate with existing systems, and how edge cases will be handled. As AI coding tools become more sophisticated and accelerate development, Kumar posits that this gap between intention and implementation is widening, making robust specification even more critical.
Introducing the Spec Review Platform
To address this, Rahul Kumar introduces easyspecs.ai, a platform designed around the concept of “Spec Review.” The core philosophy, as Kumar explains, is to “resolve the hard conversations before anyone writes code.” This approach prioritizes clarifying intent, visualizing changes, and defining success criteria upfront. The proposed workflow emphasizes a structured progression: Change → Intent → Diagram → Spec → Trust Engineering → Code.
Key Features of Spec Review
Kumar points to several standout features of this spec-centric approach:
- Spec Review: This allows for the examination of requirements before they evolve into implementation challenges.
- Document Review: It aims to foster a shared understanding of the existing system, moving away from reliance on undocumented “tribal knowledge.”
- Specs Readiness: This feature provides clarity on what elements are blocking the completion of an implementation.
- Trust Engineering: Kumar suggests this transforms requirements into verifiable claims, replacing subjective assessments with objective checks.
- Agentic Flow Design: The platform facilitates the mapping of AI agent steps, programmatic actions, and evaluation metrics prior to execution.
- Jira + Linear Integration: This integration streamlines the process from initial stakeholder requests to a well-defined, developer-ready task.
The Evolving Role of AI in Development
A significant point Rahul Kumar makes is the evolving relationship between humans, AI, and the specification document. He argues that as AI generates code at an unprecedented pace, speed without a shared understanding can lead to more rapid errors. Kumar’s central thesis is that future successes in AI-assisted development will hinge not just on code generation speed, but on the ability to correctly establish intent, context, and trust before code is produced.
“The teams that win with AI won’t necessarily be the ones generating the most code. They’ll be the ones that get intent, context and trust right before the code is generated.”
Kumar concludes by framing easyspecs.ai as a solution tackling this crucial problem, particularly relevant as AI-native development becomes more prevalent. He poses a provocative question to his audience: “Should Spec Review become the new Code Review?”
“AI can generate code incredibly fast. But speed without shared understanding simply creates faster mistakes.”
His post encourages feedback and support for easyspecs.ai on Product Hunt, underscoring his belief in the transformative potential of prioritizing specifications in the software development lifecycle.
📝 About This Content
This article is based on insights shared by Rahul Kumar on LinkedIn.
📅 Originally posted on September 11, 2026 | View original post on LinkedIn →