BDC AI Research Assistant

The BDC Reporter introduces its AI Research Assistant. Ask any question you like about the recent performance of any of the 45 public BDCs we track.

💡

This may be the most important of the BDC Reporter's Subscriber tools. As we've explained in an article, artificial intelligence (AI) has drastically changed the speed and breadth of BDC financial research. Assembling, parsing, and organizing the multitude of data points that BDCs regularly spew out - and which are critical to understanding both past and future performance - used to take analysts and investors countless hours. Thanks to AI - when properly applied - that time frame has shrunk enormously. Effectively, most everything one needs to know from the public record is available as soon as it's published. Just as importantly, all that data is universally available. No need for a dedicated terminal or subscriptions to multiple data services.

Apologies, but we've had to withdraw the BDC AI Assistant until we can resolve some technical difficulties. We are frankly amazed at how versatile the Google NotebookLM tool is that we're using can be if all the right data is input. It goes far beyond information retrieval and extends to being able to undertake sophisticated financial analysis when asked. Sometimes the AI even suggests useful analysis on its own volition. Less flexible is Google which makes the sharing of this modern day wonder with our readers as difficult as possible and susceptible - like any early days prototype - to breaking down. That's our current challenge just after, ironically enough, the key sources for every BDC we track from the two most recent quarters - ideal for answering any imaginable question about everyone's financial performance. Wew'll see if we can get this marvel back into working order. Please stand by - June 20, 2026

Good Old Days?

In the past - which means up till a few months ago - you'd have to either read all the earnings releases, investment presentations, conference call transcripts, and quarterly filings that a BDC published to get a picture of what was going on. That took time as you moved from source to source, taking notes along the way. Then you might have to wait for a financial publication or an analyst to use some part of that information to write something of interest about the BDCs you were interested in. With the benefit of hindsight, we can see the process was slow and inefficient.

Right Now

AI has changed all that. The technology is very good at gathering all the materials needed for BDC research in one place as soon as they appear and making them readily available. In our case, we use Gemini's NotebookLM, which allows us to gather up to 600 different sources on a subject under one roof and then ask... anything. One can inquire about an obscure fact, create a table comparing multiple metrics, or review word-for-word what was said about any subject of interest. The only limit is one's own level of interest.

Packaging

With that said, a great deal of manual work has to be done behind the scenes. A human "agent" has to identify which source materials are worth gathering, and which are not, and then do so in a timely manner, preferably almost immediately after publication. That same human agent also has to organize the sources in an optimal format for the AI to understand and constantly cull out-of-date material to ensure the best results. Finally, a human agent needs to regularly check that everything is working as expected. That's what the BDC Reporter is doing here.

All Set

When all the above is done right, NotebookLM and similar AI can offer BDC research as we've never known before, at little cost and at any time, day or night. That's what's on offer above when you click on the BDC Reporter's AI assistant. You can "chat" to your heart's content about the recent performance of all 45 public BDCs we cover. We will be the first to admit that the AI assistant knows more than we or anyone else ever could.

Caveats

After that big build-up, here are the key limitations you should be aware of. Due to storage limits, we retain data only for the most recent quarter and the previous period. In an ideal world, the notebook would stretch back in time to encompass every source that saw the light of day, but even AI has its limits. Then there's the occasional hiccup when the AI does not fully understand a question and offers the wrong answer, or misses some files. We don't get any "hallucinations" because the AI is working with ring-fenced data, but the output isn't always perfect.

Remember This

We offer the BDC Reporter's AI Assistant as a grand experiment. The way you've undertaken your BDC research has changed dramatically, but where this will all end up, we don't know. Expect us to make tweaks along the way. Here are some suggestions for questions to ask the AI Assistant.

WHAT'S LEFT FOR US: Frankly, we're excited about the new technology. News gathering has always been a necessary element of what we do, but never the goal. Now we can spend more time on distilling that endless supply of BDC information literally at our fingertips into analysis and conclusions that can be used to make better investment decisions. You'll find our proprietary analysis and opinions woven into the AI Assistant's responses and clearly spelled out. Rightly or wrongly - as time will tell - we believe humans can out-reason their technological assistants. Our readers will be the judge.