Financial Data, AI, and Open Source Culture

While writing this week’s piece, two different posts on X and LinkedIn caught my attention and truly inspired me.

Last week, dear Yusuf Altunbıçak released an open-source code to manage Outlook 365 from the terminal. By the end of the first week, a developer from 7,000 km away in China came along, reported some issues, and suggested changes. Time zone support, a negative day parameter, automatic login, three features in one go.

And when the responses to these change requests didn’t come quickly enough, he wrote the code to fix them himself and sent it to Yusuf for approval. People from the other side of the world took ownership of the project and contributed to it. That’s how open source works, really, though I know Yusuf’s example is a bit extreme; he must have picked a popular topic. But that’s the open-source world for you: you open a door, and you never know who might walk through.

I came across another project created with a similar mindset. I first heard about Said Sürücü’s Borsa MCP at the second Claude meetup organized by Said and Onat. Last week, he shared its usage numbers on LinkedIn, exactly 202,175. These two posts encouraged me to write this piece. It sounds technical, but at its core, it connects financial data to AI tools.

The MCP server is named “Borsa,” but that name doesn’t really do the project justice. Once you read the details, you see:

758 companies on BIST, the US stock market, over 836 investment funds, crypto, commodities, currency, KAP news. On top of that, Buffett analysis, Altman Z-Score, participation finance compliance, and connecting to Claude takes just a few minutes.

Borsacı takes it a step further: it’s a ready AI agent using Borsa MCP that can handle multiple tasks at once and has conversational memory. The setup is a bit more technical, but looking at the big picture, it’s clear where tools like this are headed.

I tried connecting Borsa MCP to Claude and creating the skill to use it, working on a few stocks. Fundamental analysis, balance sheets, profitability ratios, dividend history, the company’s story. Then technical analysis, hourly and 4-hour charts. It even provided a chart. Not a bad starting point at all for a long-term investor.

But don’t get me wrong: especially in the world of crypto and forex, with futures and high-leverage trades, the efficiency of these analyses can be debated. The decision and strategy are still very much human. This isn’t an agent, but a very useful tool for information and analysis.

So here’s the main point. What’s the common thread in these projects?

Open source. Artificial intelligence. Collective growth.

Said built a financial data infrastructure starting from Turkey. Yusuf released a command-line tool, and someone from China came along and expanded it. People unaware of each other, yet part of the same open-source culture.

AI acts as a multiplier here. Regardless of how good an analyst you are, those using these tools can look at a broader picture more quickly. But no matter how good the tool is, it’s still the human who asks the question, sets the context, and makes the decision.

And what about agents? “Let AI trade for me” is a real expectation, and users are already leaning that way.

Life is heading in that direction, or maybe it’s already there. But how exactly agents will act, where their boundaries will be drawn, and how security will be ensured are still questions waiting for answers. I’ve written about these topics before, and I’ll keep writing about them.

For now, we need to talk about the infrastructures, because assistants, agents, and much more will operate on these foundations.

Discussion question: AI is a powerful tool for financial analysis, I’ve definitely experienced that. But when agents arrive, when the system starts making decisions for you, how far can the system and agents truly represent you?