Vibe Coding: The Missing Link
Decades of layering on knowledge and improvements with tooling, open-source software, package managers, git-based systems – and more - have brought us to a place where the unimaginable will become commonplace. This is because machine learning and artificial intelligence have reached a level of maturity and capability revealed through MCP servers where every aspect of our lives can once again be transformed by the massive amounts of data that exists online in both public and private systems.
One thing was missing, however – a principled guidebook to instill order in the chaos.
One attempt to grapple with instilling order in the chaos is the AI-Driven Development Lifecycle (AI-DLC) Method Definition whitepaper by AWS (https://prod.d13rzhkk8cj2z0.amplifyapp.com/)
AI-DLC is a new mindset using AI-centric systems and tooling to support developers (and knowledge workers!!!) through every step of building complex software. One that aligns very well with DevOps, agile workflows, and cloud-native environments.
The whitepaper identifies how “existing software development methods, designed for human-driven, long-running processes, are not fully aligned with AI’s speed, flexibility, and advanced capabilities.”
So, they reimagined SDLC methods for the needs of today and came up with the AI-Driven Development Lifecycle (AI-DLC). A word of caution, this is a reimagining, a shifting of the mind, not another cloud “lift and shift” moment where we jam round pegs into square holes and call it transformation.
Remember, this non developer, motivated by this sense of transformation, followed this process and was able to go from idea to production ready SaaS app in a couple days. Imagine the velocity and value creation potential of a developer who wouldn’t have struggled with beginner hiccups like I did. Imagine the impact to code quality so that cross site and SQL injection and input validation vulnerabilities disappear from our vocabulary.
This is something I am passionate about. While at AWS as a security architect I would try to hammer home how a short list of baseline templates would eliminate large areas of risk. I’d build demos based on latest breach reports to show customers how the baselines paired with AWS security services would have mitigated those attacks. When I moved to RockITek, focused on helping emerging tech companies design and build products that would be fit for purpose in government environments, I created the tagline “we aspire to make adversaries irrelevant” because I wanted to communicate clearly that we could help others build systems and products that would increase the cyber economic cost to the adversary, impacting adversary decision making process with a high probability of reducing the risk to protected systems. At NXT1 this passion is being expressed in a platform that aspires to simplify and automate as much as possible so that every SaaS builder can have a secure by design and default place to deploy their products and focus on building their customer base and features.
Similarly, I believe this is an opportunity to reimagine development by figuring out how product requirements and user stories in markdown requirements documents can align the human / AI understanding of what needs to be built, creating “well-defined contracts.” Contracts which can also include a serious leap forward in effective shift left thinking by adding threat models, compliance and organizational risk factors into the initial requirements docs that guide AI’s responses and behaviors. Think about that for a moment…
The Open-Sourcing Adaptive Workflows for AI-Driven Development Life Cycle (AI-DLC) blog provides more context on how to reimagine AI-DLC by identifying 3 recurring challenges faced by software engineering teams and how they can be mitigated. The bottom line, AI needs to be part of the team, where clear roles and responsibility exist. It discusses ways to operationalize and embed AI-DLC principles, concluding the answer is workflow scaffolds. These are Rules or Steering customizations for AI Coding Agents to operationalize AI-DLC principles within the tools. This is reusable, realistic and an effective way to balance AI and human experience.
Recommended by LinkedIn
As I concluded in my knowledge workers vibe coding article, the real power is in the combination of human experience paired with AI. It must be an interaction where we pay attention to the logic used by our coding assistants, how they reason about a design or troubleshooting step so we can guide it on big picture thinking and correct hallucinations before they cause bugs and delays.
Specific examples were discussed in the Building with AI-DLC using Amazon Q Developer blog. Here they identify how the phases: Inception, Construction, and Operations, are a workflows to “maintain quality and control through structured milestones and transparent decision-making.”
They also provide a walkthrough to get your hands dirty and your mind transformed. AWS created an aidlc-workflows repo with the baseline tools you need to get started.
FYI, here are 2 great resources in the repo you really need to read:
If you want more information about the specific rules, you can take a look at this doc Creating project rules for use with Amazon Q Developer chat. This is the most important part. It is the baseline from which all code, automations, configurations and deployment processes flow.
Creating project rules is critical to ensuring you can achieve the desired outcome because the AI assistant (Kiro, Amazon Q, etc) will automatically use them as context for all interactions and actions within that project when generating answers. This will help remove many headwinds and reduce debugging later.
I’d recommend building a Templates folder with these already baked in, so when you start a new project, you can copy and rename the folder, so I have all your core requirements set. All you have left to do is modify any rulesets based on your goals and define the application you’re building.
Good luck and let me know how it goes!
