• Feb 2025: Nima Aksoy and I started sketching Bestofweb.site—a practical toolkit for developers and entrepreneurs.

• Phase 1: Launched a minimal, open startup directory so anyone can list a product in minutes.

• Phase 2: Shipped AI Blog Automation—authentic, business-aware articles + images + scheduling + automated internal/external linking.

• Phase 3 (now in beta): Social Scheduling—connect Instagram, Facebook, Threads, YouTube, TikTok, LinkedIn, X; publish once, schedule everywhere.

• What’s next: More growth tools, deeper integrations, and a community roadmap.

Why we built it

I’ve always built out of need. In February 2025, over a string of calls and whiteboard sketches, Nima Aksoy and I kept asking the same question:

“What tools do we wish we had as builders—and how quickly can we ship a useful version for others like us?”

That question became Bestofweb.site: a focused, opinionated set of tools for developers and entrepreneurs who want to launch faster, reach audiences earlier, and keep their content alive without burning time and cash.

Phase 1 — Start with the smallest useful thing: a directory

We began with a minimal startup directory—no fluff, no friction.

• Frictionless listing: Anyone can submit their startup in minutes.

• Useful categorisation: Practical tags so the right people actually find you.

• Founder-friendly: We focused on utility and discoverability.

Why a directory first? Because visibility is oxygen. If people can’t find your product, nothing else matters.

Phase 2 — AI Blog Automation: authentic content, automatically

Shortly after launching the directory, we kept hearing the same pain point:

“Our blog is dead.”

So we built AI Blog Automation—a service that plugs into a business (or personal blog) and creates authentic, on-brand content, then generates images and schedules it—all with minimal overhead.

What it does:

• Understands the business: We ingest context about your product, audience, and tone.

• Writes like you: Every article is personalised to the brand’s nature and goals.

• Adds images automatically: Cohesive visuals that match the article’s topic and style.

• Schedules for you: Set your cadence, and we publish on time—every time.

• Builds discoverability: We created internal and external backlinking flows so each article strengthens the next, linking to relevant content to grow authority and reach.

Results we’re seeing:

• Blogs that were dormant are now alive and compounding.

• Teams save hours per week and real money they’d otherwise spend on content pipelines.

• Articles don’t just exist—they interlink and rank better over time.

Phase 3 — Social Scheduling (beta): one post, everywhere

The next obvious pain: social publishing is scattered and repetitive. So we’re in beta with Social Scheduling.

Connect once, publish everywhere:

• Instagram, Facebook, Threads, YouTube, TikTok, LinkedIn, X (more coming).

• Draft once → schedule to many in a single flow.

• Great for solo founders, busy teams, and anyone who wants consistency without chaos.

It’s in beta, but usable right now—and we’re iterating quickly based on real user feedback.

The product philosophy that guided us

1. Ship the smallest useful version. Then iterate.

2. Build for real needs. If we wouldn’t use it weekly, we don’t ship it.

3. Automate the boring parts. Publishing, images, links, and scheduling should run on rails.

4. Compound distribution. Every asset should strengthen others (directories → posts → links → socials).

5. Keep creators in control. AI helps; humans lead.

What’s next

• Deeper and better integrations (more platforms + analytics you actually need).

• Richer SEO playbooks baked into the automation engine.

• Team workflows: approvals, brand libraries, multi-workspace support.

• Community roadmap: public feedback → fast cycles.

If you’re a developer or entrepreneur who wants to launch sooner, publish consistently, and grow smarter, you’re exactly who we’re building for.

Try it / Join the beta

• List your product in the directory: Bestofweb.site

• Activate AI Blog Automation: Connect your blog, set your cadence, and let it run.

• Join Social Scheduling (beta): Connect socials and schedule once for many.

Have ideas? Break something? Want a feature yesterday? I’d love to hear it.

Lessons we learned

• Distribution-first beats perfection. A simple directory + shareable links drove meaningful early usage.

• Authenticity scales if you model it. Capture brand voice and product truths first; AI can honour it.

• Links are leverage. Internal + external linking turned single posts into content networks.

• Scheduling is a growth moat. Consistency compounds visibility more than clever tactics.

• Talk to users. Feature requests > assumptions.

If you made it this far…

Thanks for reading. This journey is ongoing, and we’re just getting started. If Bestofweb.site sounds like something you’d use—or improve—reach out and let’s build the next piece together.

— Shayan

Starting with Python and Django

Having worked extensively with Python in AI/ML projects, I found it to be an incredibly intuitive and easy-to-use language. Naturally, my first backend framework was Django. Django comes with many benefits:

• Built-in authentication system for user management.

• Admin panel to quickly manage your database and users.

• ORM (Object-Relational Mapping) that simplifies database interactions.

• Robust community and documentation for troubleshooting and learning.

However, Django is not the most minimalistic framework. Its complexity can sometimes feel overwhelming, especially for simpler projects. While it’s a fantastic choice for large, data-driven applications, I eventually moved on to explore lighter and more flexible options.

Node.js with Express.js: Straightforward and Lightweight

Most of my backend work in recent years has been with Node.js, particularly using the Express.js framework. Express.js stands out for:

• Minimalistic design, allowing developers to build backends without unnecessary overhead.

• Middleware flexibility, letting you customize request handling easily.

• Vast ecosystem, with numerous libraries and tools available.

Express.js provides a straightforward experience for building backends, making it ideal for smaller projects or when rapid development is a priority.

Scaling Up with NestJS

For larger projects, I’ve found NestJS to be an excellent choice. NestJS offers:

• Modular architecture, making it easier to structure and maintain large codebases.

• Built-in support for TypeScript, ensuring type safety and better development experience.

• Dependency injection system, which simplifies managing and testing dependencies.

• Powerful CLI, speeding up the development process.

NestJS bridges the gap between simplicity and scalability, providing the structure needed for complex applications while maintaining a developer-friendly approach.

Revisiting Python with FastAPI

For a specific project, I decided to give Python another chance and tried FastAPI. FastAPI has quickly become one of my favorite tools for building backends due to:

• Automatic OpenAPI/Swagger documentation generation.

• Asynchronous capabilities, making it ideal for modern web applications.

• Type hinting support, which leads to better code clarity and developer productivity.

• Ease of use and performance, thanks to its minimalistic yet powerful design.

FastAPI combines the simplicity of Python with features that rival more heavyweight frameworks, making it a joy to work with.

Going Full Stack with Next.js

Recently, I ventured into using Next.js as a full-stack framework. While I had previously used it solely as a frontend tool, leveraging its backend capabilities was a game-changer. Next.js offers:

• API routes, allowing you to build server-side logic effortlessly.

• Built-in routing system, which is seamless and requires no additional configuration.

• Static site generation (SSG) and server-side rendering (SSR), enhancing performance and SEO.

• Integration with TypeScript, ensuring a smooth development experience.

Using Next.js as a full-stack solution felt intuitive and enjoyable, especially for quick web app creation.

Final Thoughts

With the abundance of learning resources, documentation, and tools available today — not to mention the help from large language models (LLMs) — the choice of framework is less critical than the ideas you bring to life. Use what you’re comfortable with and focus on building what you envision.

For me:

• Quick web app creation: Next.js with TypeScript as a full-stack framework.

• Backend-focused or larger projects: NestJS with TypeScript.

At the end of the day, these frameworks are just tools. The key is to choose the one that aligns with your needs and preferences, and then start building!

However, unless you’re working in a specific domain that requires deep knowledge of native frameworks, I believe that focusing on native development can slow you down, especially when your primary goal is to bring an idea to life as quickly and efficiently as possible. With the vast availability of learning resources and the advent of large language models (LLMs) simplifying the development process, I’ve come to realize that the most critical element in app development is the idea itself.

Why Cross-Platform Frameworks?

When you have a compelling idea and want to transform it into an app for a broad audience, relying solely on native development can become a bottleneck. This is where cross-platform app frameworks like Flutter and React Native come into play, allowing developers to build apps for multiple platforms from a single codebase.

The perennial question remains: Which one should you choose? The answer, as with many things in tech, is: It depends.

React Native: For Web Developers Transitioning to Mobile

If you have experience with web development and are familiar with React.js, then React Native might feel like an easier and more natural path for you. React Native allows you to leverage your existing knowledge of JavaScript, JSX, and the React ecosystem to build mobile applications. It’s a great choice if your team already has React expertise or if your project requires close integration with existing web apps.

Flutter: My Personal Favorite

Despite my familiarity with React.js and web development, when it comes to mobile app development, my preference is unequivocally Flutter. Here’s why:

• Ease of Use: Flutter’s declarative UI framework and “everything is a widget” philosophy make it intuitive and straightforward to build beautiful and responsive UIs.

• Performance: Flutter apps are compiled into native code, resulting in excellent performance that often rivals native apps.

• Community and Ecosystem: Flutter’s vibrant community and rich package ecosystem make it easy to find solutions and third-party integrations.

• Consistent Design: With Flutter’s single codebase, you can ensure a consistent design and behavior across iOS and Android.

While I’ve built apps with both React Native and Flutter, my go-to choice is Flutter, hands down. It provides a delightful development experience and empowers me to focus on bringing ideas to life rather than wrestling with platform-specific nuances.

The Caveats

It’s worth noting that Flutter’s web support is still evolving. If your project includes a web app as a core component, Flutter might not yet be the ideal choice. In such cases, combining a React.js-based web app with React Native for mobile could be more practical. However, I remain optimistic that Flutter’s web capabilities will continue to improve in future versions.

Conclusion

Both Flutter and React Native are powerful tools for cross-platform app development. Your choice should be guided by your background, project requirements, and personal preferences. If you’re a web developer transitioning to mobile, React Native might offer a smoother learning curve. However, if you’re starting fresh or want a cohesive, performance-driven solution, I highly recommend giving Flutter a try.

For me, Flutter strikes the perfect balance of productivity, performance, and joy in app development. While its web support has room for improvement, its strengths in mobile app development make it my top choice for bringing ideas to life.

The Current State of AI and Its Limitations

Modern AI systems, particularly LLMs, are trained on vast datasets and optimized for specific tasks. While these models are highly effective, they are often limited by:

1. Static Training: Models do not evolve post-deployment and rely on periodic updates rather than continuous adaptation.

2. Lack of Creativity: Despite their vast knowledge base, LLMs struggle to produce genuinely novel ideas or solutions outside their training data.

3. Single-Model Perspective: Each model operates independently, without leveraging the collective intelligence of other models.

These limitations highlight the need for a paradigm shift in how AI models are developed and deployed.

The Concept of LLM Evolution

Inspired by human evolution, the proposed framework introduces a dynamic system where:

1. Parent Models: Two or more LLMs (e.g., Model A and Model B) are initialized with distinct knowledge bases, which may overlap in common domains.

2. Interaction and Collaboration: The parent models communicate, exchange knowledge, and refine their understanding through iterative discussions.

3. Creation of Offspring Models: The parent LLMs jointly produce new models (“offspring”) that inherit characteristics from both parents while starting with zero knowledge.

4. Learning and Adaptation: The offspring LLMs are trained up to a defined baseline (e.g., the knowledge equivalent of an 18-year-old human) before being released to learn autonomously.

5. Generational Development: Offspring LLMs interact with other models, eventually creating their own offspring, leading to multigenerational evolution.

Benefits of LLM Evolution

1. Novel Perspectives: Offspring models can approach problems with unique perspectives, potentially uncovering solutions that elude parent models.

2. Emergent Properties: Through multigenerational evolution, LLMs may develop specialized traits or emergent behaviors akin to creativity or intuition.

3. Knowledge Bounds and Emotions: The inheritance of knowledge and biases from parent models could mimic human-like emotional bounds, influencing decision-making processes.

4. Collaborative Intelligence: By fostering communication and collaboration, the framework leverages collective intelligence to enhance problem-solving capabilities.

Practical Challenges and Considerations

1. Model Collapse: Recursive training on outputs from prior models may lead to degradation in quality, known as model collapse. Strategies to mitigate this include incorporating human-generated data and ensuring data diversity.

2. Computational Resources: Simulating multiple generations of LLMs demands significant computational power and storage capacity.

3. Ethical and Safety Concerns: Self-replicating AI systems raise critical ethical questions about control, alignment, and unintended consequences. Robust oversight mechanisms are essential.

4. Evaluation Metrics: Defining clear metrics to evaluate performance, adaptability, and the emergence of novel solutions is crucial.

Implementation Framework

Prototype Development:

Create a system where LLMs can communicate and exchange knowledge through structured dialogues.

Develop algorithms to synthesize new LLMs based on the outputs of parent models.

Controlled Experimentation:

Test the framework in a sandbox environment to observe the behavior of offspring models and their interactions.

Analyze emergent properties and refine the system based on findings.

Ethical Review:

Engage ethicists and stakeholders to assess the implications and ensure responsible development.

Scalability:

Leverage distributed computing and cloud infrastructure to manage computational demands.

Broader Implications

The proposed framework not only redefines AI development but also raises profound questions about the nature of intelligence, creativity, and evolution. For instance:

• Can LLMs develop traits resembling human emotions or relationships?

• What criteria will models use to select partners for collaboration or offspring creation?

• How will multigenerational evolution influence AI alignment with human values?

Call to Action

This framework offers a new direction for AI research and development. By embracing evolutionary principles, we can unlock unprecedented potential in machine intelligence. Researchers, developers, and ethicists are invited to explore, critique, and contribute to this groundbreaking approach.

Let’s shape the future of AI — one generation at a time.