DEV Community: Bartolomeo Hart

My 48h SaaS Challenge

Bartolomeo Hart — Thu, 27 Mar 2025 18:26:51 +0000

Just wrapped up a challenge to build a SaaS as quickly as possible 🚀

The goal?

Create a working SaaS product that analyzes crypto content, solo from scratch, within the timespan of 2 days!

Here's how it went...

I spent the first 6 hours researching, planning and designing what to build and how to build it. With crypto being my focus area, I wanted something that could extract insights from videos and provide actionable data.

The frontend work consumed about 10 hours, building a responsive Next.js website with landing page and dashboard.

I was able to build a complete frontend containing:
• A pricing overview page
• Checkout flow
• Dark/light mode with persistence
• Full legal documentation

And much more, even dark/light mode. It's actually cool that it turned out this way with limited time!

The dashboard uses Chart.js for visualizations and lets users track price targets, follow event timelines, and filter through raw data, which can be exported as .csv and used for ones own purposes.

Backend development took roughly 6 hours, setting up FastAPI, YouTube caption extraction, and database integration. While I've built Flask and Django backends before, this was my first FastAPI project.

And that's where things got tricky. 😅

The FastAPI with background tasks became a nightmare, concurrent operations kept hanging. With only 4 hours left to debug, I couldn't solve the fundamental issues with the AI-generated boilerplate code.

Payment integration was another unexpected challenge. Cryptomus rejected my application twice, first for an incomplete offer (fair), then because "we do not work with this category of projects." Despite checking their guidelines thoroughly, I never found any mention of restrictions on crypto+AI projects.

The biggest lessons I'm taking away:

Don't experiment with new tech stacks during tight deadlines, especially AI-generated ones. Stick with what you know or learn new tools beforehand.

Payment integration needs more planning than I anticipated. Having a complete, polished product ready before applying to payment providers is crucial.

I spent too much time (again🫠) perfecting the frontend while the backend needed more attention, a classic mistake when racing against the clock.

Despite the challenges, I'm not abandoning CryptoGist. It's actually a tool I genuinely need myself! I'll keep developing it, try different payment processors like Coinbase Commerce, or maybe make it open source.

Checkout the website to see the final result: https://www.cryptogist.xyz

What's next?

This 48-hour sprint taught me what's possible with modern tools, and eventhough there where some pitfalls, I believe that challenges such as these will allow you to become better and better with each iteration, allowing you to slowly setup a framework for success and a structured approach to quickly get stuff off the ground.

I like to try out stuff and build cool things, give me a follow on x.com if you want to see more of my failed attempts (and successes)👋

Get Started with Wishlists on Github

Bartolomeo Hart — Thu, 27 Feb 2025 22:40:19 +0000

Ever wondered what happens to all those brilliant feature ideas and tool concepts floating around in peoples minds and conversations? They often remain just that, ideas.

That's why I created Wishlist, a GitHub-based ecosystem to transform scattered ideas into structured proposals that can eventually become "Awesome" projects.

🧞 How it works

Wishlist is essentially the "Pre-Awesome" stage in the innovation pipeline:

Web → Scattered Wishes → Wishlist → Feature/Tool → Implementation → Awesome

The formal process is straightforward:

Fork/clone the wish template from the repo
Fill in the structured template with your idea
Submit a pull request to the Wishlist
Community members discuss, refine, and validate the proposal
Developers can fund it, build it, or collaborate on implementation

Using an RFC-inspired structure with GitHub Repos for lists and Gists for individual wishes, every proposal is version-controlled, shareable, and ready for community input.

📜 Example wish

The first wish has just been submitted (with more wishes in the works)

AI-Powered Proposal Builder

An AI tool that converts simple ideas into structured technical proposals (wishes) in milisecond🤖

This tool would help:

Newcomers without technical experience share great idea's
Busy devs needing quick boilerplate buildplans
Anyone with a good idea but lacking technical writing skills

The process is beautifully simple:

Input your raw idea
AI structures it properly
Get a complete wish proposal
Submit or refine

No more staring at blank templates wondering where to start!

Interested in building this idea?

Have a look at the wish example or read our manifesto and start adding your own brilliant ideas

Don't want to add your idea just yet?

Save and star⭐ our repo for later github.com/unameit10000000/wishlist

Thanks for helping in bridging the gap between brilliant ideas and implementation!

Wishlist: Transform Idea’s into Awesome Stuff

Bartolomeo Hart — Wed, 26 Feb 2025 00:08:51 +0000

The Awesome Phenomenon 🕶️

The 'Awesome' brand has become one of the biggest household names in software without even being a company. As a developer, you've undoubtedly encountered it whether through the official Awesome repository, Font Awesome, or one of the countless Awesome-marked resources floating around GitHub and the wider web.

If you haven't heard about it,

Awesome is essentially a GitHub repository acting as a central hub that links to other carefully curated lists following a specific format. These lists link to cool opensource projects related to everything surrounding tech: programming languages, libraries, protocols, design, gaming, AI, and much more. It's become the go-to resource for developers searching for high-quality, community approved content.

Awesome gained serious momentum around 2014, with early repositories like Vinta Chen's "awesome-python" and Sindre Sorhus's central "awesome" list establishing what would become one of the greatest movements in the opensource community.

Today, both repos are well over 200k stars, making them among the most-starred repositories on GitHub. If you've built something awesome, the Awesome repo is where you'd showcase your creation.

However,

What about all of us with brilliant ideas for tools or features that don't exist yet? Where do we go?

A Creator's Dilemma 🪦

If you're anything like me, your mind is constantly occupied with new and sometimes crazy ideas. Days without thinking about new apps, features, or tools are a rare exception.

The problem I face is the impossibility of keeping up with my own thoughts and building everything myself. While the reasons for this may vary for others, for me, it's simply a lack of skills, time, and resources.

Questions I've been asking myself are: Where do I start? Who’s going to take my idea seriously? How can I quickly expose my idea to a lot of people to get help, validation, and hopefully, build it? This struggle has been a major personal frustration I've carried for the past decade in software, and I'm certain I am not the only one.

But these roadblocks have never stopped me from trying, and they shouldn't stop you either. So what's the solution?

Enter Wishlist 📜

Wishlist is The 'Pre-Awesome' version of the Awesome repo. This is a place to propose and share innovative tool and feature ideas that don't exist yet, but could be a gamechanger for developers, creators, and enthusiasts. If you've ever dreamed about opensource features or tools that could make life or work easier, this is where you can dream big and add your vision to a growing collection of possibilities.

From Wish to Awesome 💫

Wishlist offers a fresh approach to transforming ideas into reality. Rather than letting great concepts remain scattered across forums or locked in our heads, it provides a standardized platform for two main purposes: proposing entirely new tools that might not exist yet, and suggesting innovative features for existing projects.

The idea is simple:

Web → Scattered Wishes → Wishlist → Feature/Tool → Implementation → Awesome

RFC-Inspired Structure 🦋

What makes Wishlist particularly powerful is its RFC-inspired approach to structuring proposals. Similar to how many established projects manage their feature requests through Request for Comments (RFC) processes, Wishlist adopts a formalized approach to documenting and discussing ideas.

The key difference? Wishlist leverages both GitHub Repo's and Gists as the primary medium. Repo's for lists and Gists for wishes.

Gists, for example, are perfect for casting wishes because they're:

Version-controlled
Easily shareable with unique URLs
Very lightweight and don't require full repository creation
Allow for comments and community discussion encapsulated within wish proposals

This structure creates natural compatibility with projects that already implement RFC processes, making it easier to graduate a wish into a formal proposal for an existing project. The standardized format also ensures that all crucial aspects of an idea are thoroughly considered and documented.

Pretty awesome, right?

Explore Wishlist 🌟

Wishlist is a space where innovative ideas can take root before becoming reality, waiting for the right developers and resources to bring them to life!

So, if you:

Got a groundbreaking idea that could revolutionize how developers work?
Or thought of a feature that would make a platform even better
Are looking to collaborate with like-minded people
Are looking to fund your opensource project
Want to contribute to meaningfull projects
Or want to explore what others are dreaming up

Than, make sure to Star or Fork the Wishlist GitHub Repository

Happy Wishing! 📜✨

AI is our New Drug Epidemic!💊

Bartolomeo Hart — Thu, 23 Jan 2025 19:13:05 +0000

The similarities between artificial intelligence and drugs aren't what most people expect, let alone think about.

The Path Of Least Resistance

Drug transactions🤝

Any human and animal behaviour is fundamentally just a transaction within brains where effort is traded for rewards.

Drugs exploits this wired human behaviour, transforming it into a vulnerability, offering instant gratification for actions that require minimal friction. Users trade their longterm wellbeing for immediate pleasure, damaging not only themselves but also our society in the process. At the same time, traders and traffickers are going up and beyond to provide access.

Learning from past failures

Decades of drug prevention efforts have shown that neither strict enforcement nor complete legalization offers perfect solutions. The “War on Drugs” demonstrated that just banning drugs doesn’t fix the real problems, and in some cases made things worse. Drug users and gangs always find new ways to keep going.

Why? Because using, trading and trafficing is stupidly easy! The barrier to entry is low, while the shortterm individual rewards are high.

AI crime transactions🤑

This pattern isn't unique to drug use, trading or trafficking but also other various addictive behaviors, and AI usage is following an alarmingly similar trajectory. AI is becomming THE favorite ingredient in a cybercriminal's cookbook. The parallels are:

Low barrier to entry due to low cost and required technical knowledge
Minimum effort due to ease of use, scaling and automation
High reward potential
Global reach

There's just one slight problem, or actually, two. First, borders in the digital realm are blurry and far easier to bypass. Second, it might even be easier than dealing drugs. Think about it: we live in an age where 12 year olds are rug pulling adults using memecoins and probably know more about technology than 70-80% of the population.

So what will stop people from cooking some late night snhacks using AI?🍜 Good question…

The Path Forward

The hard truth

History shows us that preventive measures against drugs have largely failed. The same challenge faces AI because we cannot control what we cannot see.

Possible solution(s)

In my opinion, the solution leans towards full transparency rather than prohibition. Just as drug abuse flourishes in shadows, AI misuse will thrive in obscurity.
We need to start asking more question, collectively, and more frequently.
Utilizing tools and processes such as blockchain and transaction identification.
More importantly, we should be very cautious.

Caution

While organized crime poses an obvious threat, we must be equally cautious about corporations, and governments who might exploit AI's capabilities behind closed doors.

Rules can be bent and sometimes are, benefiting those who create them, and whoever makes the rules has the most power.

In the realm of AI, this power dynamic becomes even more critical, where those in power will be able to control the future and every digital transaction and interaction.

"Be cautious who you give power to" - John McAfee

https://linktr.ee/bartolomeohart

Building an AI Agent Is Easy! Here’s Proof..

Bartolomeo Hart — Tue, 21 Jan 2025 22:05:59 +0000

I’ve created a simple AI agent that shows how anyone with minimal Python knowledge can get started. Let me show you how straightforward it can be!

(Note: Full code example provided below)

4 Core Components🎯

The Brain (AI Model Connection)

At its core, an AI agent just needs to talk to a language model. Here’s how you establish a connection using the Anthropic API:

def __init__(self, anthropic_api_key):
    self.anthropic_client = anthropic.Client(api_key=anthropic_api_key)
    self.models = {
        "haiku": "claude-3-haiku-20240307",
        "sonnet": "claude-3-sonnet-20240229",
    }

The Memory (Action Management)

# Register single actions
self.actions = {
    "search_wikipedia": self.search_wikipedia,
    "calculate": self.calculate,
    "get_weather": self.get_weather,
    "get_crypto": self.get_crypto_data,
}

# Register automated actions
self.automated_actions = {
    "crypto_tracking": self.start_crypto_scheduler
}

The Instructions (System Prompt)

Here’s where the magic happens. Just tell your AI what it can do:

self.system_prompt = """You are an AI assistant that can perform both automated and one-time actions.
Response Format:
- For automation: 'AUTOMATE: action_name'
- For one-time actions: 'ACTION: action_name: parameter'
- For general queries: Respond naturally
"""

The Executor (Command Processing)

Simply watch for commands and execute them:

# Check for automation command
automate_match = re.search(r'AUTOMATE: (\w+)', assistant_message)
if automate_match:
    action_name = automate_match.group(1)
    result = self.automated_actions[action_name]()

Piecing It All Together 🔄

Let’s see what this could look like:

# Init Anthropic
self.anthropic_client = anthropic.Client(api_key=anthropic_api_key)
self.models = {
    "haiku": "claude-3-haiku-20240307",
    "sonnet": "claude-3-sonnet-20240229",
}

# Create system prompt
self.system_prompt = """You are an AI assistant that ...more"""

# Create automated action
def start_crypto_scheduler(self):
    if not self.scheduler_running:
        self.scheduler_running = True

        # Schedule task for every 10 seconds
        schedule.every(10).seconds.do(self.save_crypto_data)

        # Initial run
        self.save_crypto_data()

        # Start scheduler in separate thread
        scheduler_thread = threading.Thread(target=self.run_scheduler)
        scheduler_thread.daemon = True
        scheduler_thread.start()

# Register action
self.automated_actions = {
    "crypto_tracking": self.start_crypto_scheduler
}

# Excute
def execute(self, user_input):

    # User input : "Can you monitor crypto prices?"
    self.messages.append({"role": "user", "content": user_input})
    conversation_history = [msg for msg in self.messages if msg['role'] != 'system']

    selected_model = self.models.get(model_choice, self.models["haiku"])

    # Get AI response
    response = self.anthropic_client.messages.create(
        model=selected_model,
        messages=conversation_history,
        system=self.system_prompt,
        max_tokens=1000,
        temperature=0.2,
        top_p=1.0,
    )

    assistant_message = response.content[0].text
    self.messages.append({"role": "assistant", "content": assistant_message})

    # Check for automation command
    automate_match = re.search(r'AUTOMATE: (\w+)', assistant_message)
    if automate_match:
        action_name = automate_match.group(1)
        if action_name in self.automated_actions:
            result = self.automated_actions[action_name]()
            return f"Automated task started: {result}"
        return f"Error: Unknown automated action {action_name}"

That’s it! A few simple steps to go from user input to automated task execution.

Real-World Example🚀

Want to try it yourself? Check out the full project code and get started within minutes: https://github.com/unameit10000000/reallysimpleagent

Takeaway 💡

Building an AI agent does not need to be complex. With just basic Python knowledge and a clear structure, you can start building your own agents incrementally and create something powerfull.