Rating Blog Posts With The Help of AI

I just recently published The End of GitHub (As We Know It). I documented the writing process on X

Using all the tools has helped me structure my thoughts into a blog post. Also I think that the post is over the top and might not resonate with all audiences. This is when I thought: Well when I use AI to help me in creating a post, what about getting insights from various personas giving feedback on the post.

This is exactly what I did and I like the feedback from all the personas I suggested.

Here is the prompt I used in Claude Opus 4

Please do a couple of reviews about this article 

https://www.dariuszparys.com/p/0747340a-0faf-44d0-ad4b-df6991aa183b/?member_status=free

and switch into personas that will read this article. Each Persona will rate this article based on their role and write a feedback comment from their perspective. Here are the personas to use

- CTO
- Senior Developer, no AI Coding Experience at all
- Junior Developer, no AI Coding Experience at all
- Web Developer, used AI coding tools like V0 and Cursor
- Backend Developer, used AI Coding Tools like Claude Code
- Aspiring AI Engineer, want to go into AI has experience with VSCode and GitHub Copilot
- CTO of GitHub
- GitHub Advocate

and this is the Feedback I received

Feedback from AI Personas

CTO Perspective

Rating: 8/10

"This article raises critical strategic questions about our development infrastructure investments. The author makes a compelling case about the shift from human-centric to AI-centric workflows. While I appreciate the vision, the timeline feels aggressive. Most enterprises still have substantial human teams, and the transition costs of abandoning established workflows could be significant. The security and compliance section needs more depth - audit trails are just the beginning of what we'd need. However, the core insight is valuable: we should start preparing for this shift now, even if full adoption is years away. The greenfield vs brownfield distinction is particularly useful for planning our approach."

Senior Developer (No AI Coding Experience)

Rating: 5/10

"This feels like another 'everything will be automated' piece that ignores the complexity of real software development. The author dismisses GitHub's collaboration tools too quickly - Issues aren't just for tracking work, they're institutional knowledge. PRs aren't just checkpoints, they're where we share context and mentor junior developers. Yes, AI can write code, but who debugs the AI's bugs? Who explains why certain architectural decisions were made? The article assumes AI agents understand business context perfectly, which hasn't been my experience with any automation tool. It's thought-provoking but oversimplifies the human elements that make software development successful."

Junior Developer (No AI Coding Experience)

Rating: 4/10

"This article terrifies me. If AI agents are going to handle everything from requirements to deployment, where does that leave developers like me? The author barely addresses the human side of this transition. How do I learn if there are no PR discussions? How do I understand architectural decisions if everything is just markdown files processed by AI? While the technical vision might be sound, it feels like it's written by someone who's forgotten what it's like to be learning. I wish there was more discussion about how junior developers fit into this AI-driven future."

Web Developer (Used V0 and Cursor)

Rating: 7/10

"Having used V0 and Cursor, I can see where the author is coming from. These tools already blur the line between human and AI development. The spec-driven approach makes sense - I've wasted hours translating requirements into issues that could have just been markdown files. However, the article underestimates the value of visual project management and collaborative discussions. Sometimes a Kanban board helps me think through priorities in ways a folder of markdown files doesn't. The future probably lies somewhere in between - AI handling the grunt work while humans focus on creative problem-solving and stakeholder communication."

Backend Developer (Used Claude Code)

Rating: 9/10

"Spot on analysis. Claude Code has already shown me how AI agents prefer working with files over APIs. The BMAD and PRP examples are exactly what I've been experimenting with - everything as code, specs driving development. The author correctly identifies that GitHub's value proposition shifts from collaboration to infrastructure. My only critique: the article could explore hybrid approaches more. In my experience, AI agents excel at implementation but still need human guidance on architecture and business logic. The future isn't replacing developers but changing how we work."

Aspiring AI Engineer (VSCode + GitHub Copilot)

Rating: 10/10

"This is exactly the future I'm preparing for! The article perfectly captures why traditional development workflows feel outdated when working with AI. I've been frustrated by the disconnect between Copilot's capabilities and GitHub's UI-heavy approach. The frameworks mentioned (BMAD, PRP, Kiro) are going straight to my research list. The vision of repositories as the single source of truth for everything - not just code - is brilliant. This isn't about replacing developers; it's about evolving our role from code writers to AI orchestrators. The security considerations show the author isn't just hype-driven but thinking practically about implementation."

CTO of GitHub

Rating: 6/10

"The article raises valid points about evolving developer workflows, but oversimplifies GitHub's value proposition. Yes, AI agents work with files, but human developers still need collaboration tools - and they will for the foreseeable future. The author dismisses our UI too quickly without considering that most development still involves human teams. Our Copilot initiatives aren't 'polishing an obsolete system' - they're meeting developers where they are today while building for tomorrow. The infrastructure point is well-taken, and we're already investing heavily there. But GitHub has always been about more than just git hosting - it's about community, learning, and collaboration. Those human needs won't disappear, even if the tools evolve."

GitHub Advocate

Rating: 5/10

"While technically interesting, this article misses the heart of what makes GitHub special - the community. Open source isn't just about code; it's about people collaborating, learning, and building together. Issues aren't just task tracking - they're where newcomers ask questions and experienced developers share knowledge. The author's vision of autonomous AI agents committing directly to repos ignores how open source projects actually work. Who reviews contributions? Who ensures code quality and project direction? The human element isn't overhead to be optimized away - it's the foundation of sustainable software development. Yes, AI will change our workflows, but GitHub's collaborative features will evolve, not disappear."

ReIterating with this Feedback

This technique might be a good one for writers to get some hints on the bias your content has. Mine was clearly not towards GitHub, also I really love GitHub. I was just questioning myself where everything is heading.

I didn't take this feedback to reiterate over my post. I was just too lazy to restructure and rewrite parts of it again to make it more pleasant for all audiences. In the end this is my thinking and I'm fine with it.