Most organizations today see GenAI as a tool to write more code faster. While this provides short-term gains, the full potential of generative AI hasn’t been realized. As GenAI evolves, new capabilities are beginning to emerge.

We explored whether an LLM could act as an API design coach. The results were impactful.

AI-in-the-loop is the best way to describe this approach. Experts leverage prompts and LLM automation to accelerate their API design efforts for critical business systems. Let’s explore what has changed that allows GenAI to assist in the API design process. We will then explore the ADDR process and how GenAI can help.

LLMs are now ready for design-first coaching

Early exploration of API models weren’t ready to coach someone through an API design process.

Skeptical? So were we. We tried multiple times to leverage LLMs to drive design-first coaching. It just didn’t work - until recently:

  • Context windows were previously too limiting. Now, better prompt engineering can overcome this with recent AI models. This results in reduced hallucinations while avoiding poor or incomplete API designs.
  • Former LLM models couldn’t break down complex workflows. We now have advanced models that can decompose complex problems into an executable plan. The plan guides practitioners effectively through the process. Response outputs were limited to text-based formats.
  • Recent AI models have added new ways to generate output formats than in the past. This reduces the manual work required for generating artifacts, incorporating feedback, and delivering a great API specification.

What about vibe coding?

Vibe Coding emerged in early 2025 as a new development style that refers to an intent-based approach to development that uses a prompt, run, fix methodology. It is commonly used for rapid prototyping, creating disposable tools, and non-critical applications.

Business Critical AI refers to the use of AI-augmented engineering that uses a spec, prompt, test, human review methodology to produce reliable, deterministic code for critical systems. It is commonly used when reliability, explainability, and guardrail enforcement is paramount.

In short:

💡 Vibe Coding help organizations find the right idea fast.

📈 Business Critical AI helps to scale the idea safely.

AI-assisted ADDR speeds the process of transforming insights into business-critical systems.

Get the report and get started

Reduce API design from hours to minutes with LLMs as your assistant. Download our free 78-page report on AI-Assisted API design.

This is our definitive guide to using Large Language Models (LLMs) as an API Design Coach, not just a code generator using tools such as Claude Code.

Don’t rush your code out the door and create more technical debt.

Reduce your design efforts by 60% or more, without sacrificing a great developer experience.

You stay in charge throughout the whole design process. We refer to this as “AI in the loop”, rather than “human in the loop”, to ensure that your insights are first in the design process.

Download this free 78-page report on AI-Assisted API design.

Note: The report requires an email address. We only use this email address to notify you of updates and related resources. We won’t sell your details and you can unsubscribe at any time.