"Agent State-of-Play 2025: Who's Building Autonomous Investors?" Welcome, and fasten your seatbelts, because we're about to embark on a deep dive into the fast-paced, high-stakes world of autonomous investors. We're talking about artificial intelligence with a license to trade. We've all heard of self-driving cars, but what about self-driving money? Imagine a world where algorithms manage your 401k, trade stocks, and optimize your portfolio. But here's the thing – that world is already here, and it's evolving faster than you can say "Wall Street." Today, we're exploring the key players in this industry, the open-source platforms that are transforming the market as we know it, and the capabilities and gaps that are shaping the future of autonomous investing. First off, let's meet the agents. These are persistent loops that observe, reason, and act, with tool API access and a memory store. They're the proverbial brains behind the operation, constantly analyzing market data, making decisions, and taking action. One of the standout performers in this space is FinGPT. It's like the LeBron James of AI investing. This language-learning model is fine-tuned on an astronomical amount of market data, with plug-ins for parsing stocks, crypto, and even Federal Reserve speeches. It's been dubbed the future of financial analysis – and with over 21k stars on GitHub, it's clear to see why. But FinGPT isn't alone in this game. There's also BloombergGPT, a behemoth of a model with 50 billion parameters, exclusive to Bloomberg terminals. Microsoft's AutoGen, an orchestration framework that manages multiple agents, is another big player. And let's not forget Workday Agents, proof that AI can handle payroll and expense workflows, and pass SOX audits. These agents are already performing tasks that were unthinkable just a few years ago. Imagine having a research copilot that can scrape SEC-EDGAR, analyze sentiment, and output code. Or a trade planner that uses historical regimes to choose a strategy. Or an execution agent that chooses the best broker or venue for a trade. But as powerful as these agents are, there are still gaps in their capabilities. One of the biggest challenges is latency. Most language-learning models are still hosted in the cloud, which means there's a delay of 200 to 600 milliseconds. That might not sound like much, but in the world of trading, it's an eternity. In the immortal words of Elon Musk, "The market is just another physics engine—feed it faster, iterate bigger." In other words, there's a huge opportunity for platforms that can push small-form transformers to the edge, reducing latency to less than 5 milliseconds. There's also room for improvement when it comes to compliance. Few platforms have policy-as-code gates, which means there's still a lot of lawyer overhead. And while some agents are beginning to self-retrain daily on new profit and loss data, this is a feature that could be implemented more widely. So where does this leave us? Well, it's clear that autonomous investing isn't just a pipe dream – it's a reality. And while there are still challenges to overcome, the potential for growth is enormous. As these agents become more sophisticated, more efficient, and more accessible, we can expect to see a seismic shift in the financial industry. But more than that, the rise of autonomous investors is a testament to the power of technology. It's a reminder that we're living in an era of unprecedented innovation, where the only limit is our imagination. So, whether you're an investor, a developer, or just someone who's curious about the future, it's an exciting time to be alive. And with that, we've reached the end of our journey. Until next time, stay curious, stay invested, and remember – the market may be another physics engine, but it's also a playground for the mind.