The FirstBrands disaster is now behind us and we have all seen the email below, but what do it mean? What are factoring facilities? And what is going on here? 1) Factoring Basics 2) Dummy Example 3) Common Factoring Structures 4) Reasons for Factoring 5) The infamous Weil email explained

This is for my small cap investors out there: The lower you go in market cap, the more dispersion you see between sell-side consensus and reality. Bringing AI into the picture introduces some law of large numbers into the picture. 4 analysts predicting earnings? Not accurate. But a model taking in every available input? Much better. The sell side has already been incorporating AI into their reports. It’s about time we see the information from their side. I also will be curious to see how much dispersion there is across models, especially if different submissions rely on the same LLM

Influencers: “yeah, my job involves actually showing my face and talking to my audience, and it’s not like AI is ever going to give you a video as good as mine” AI:

Last chance to sign up ‼️‼️ Everyone who does will get free resources including primers, premium Pari Passu write-ups, and recorded lessons. + chance to win large discounts and a fully free program!

Historically, a key co-op objective was to prevent non-pro rata treatment of creditors Medical Solutions shocked the world and instead of prohibiting non-pro rata treatment, it promoted it through tiered structures In our coverage today, we dive in the 4 co-op tiers, the tranche sizes, the economics realized in the LME, and much more Read now, link below

meta's ai agent is 3 weeks old and the reports already include... – a columnist says it synced 187,000+ rows of his mac messages after he declined – a user says it shared his pickup location with a marketplace buyer – amazon blocked it from shopping meta's answer: the messages sync was opt-in The whole point of this autonomous technology is that people have to trust it in order to adopt it. Focusing on a trust-first experience will be absolutely key

This perfectly illustrates Dylan’s point about GPUs increasing in value over time According to Dylan, the cost of intelligence has collapsed drastically. Tokens keep getting cheaper because the same level of intelligence now comes from smaller models, which means far more tokens served per GPU. The cost per token falls, but it falls less than the increase in the number of tokens coming out. So total dollars generated by a single GPU goes up. When the dollars out of a GPU rise, the price of that compute rises with it. Therefore, the cost of compute is not falling. For an existing GPU, it is actually increasing.

Our second investment ever: Constellation Lab, the Future of AI! Running a research firm means hearing pitches from a lot of new companies. Most aren't worth your time, so you never hear about them. Some are great, and we're happy to partner with them. And once or twice a year, we find one that's truly unique, the kind we want to take a stake in and work with over the long term. Constellation is in that last group. So what do they do? As frontier AI models become commodities, a firm's real edge is its own knowledge: how its best people think and make judgment calls. Constellation helps firms turn that judgment into assets they own, including proprietary context graphs, tools, and fine-tuned models. Your expertise stays in-house, and your inference costs stay predictable. They don't just sell software. Constellation sends forward-deployed engineers, with backgrounds at firms like Citadel, AllianceBernstein, and J.P. Morgan, to work alongside your team and build exactly what you need. Palantir, but for financial services and law firms that aren't buying hundreds of millions in GPUs. I am excited because most firms buy expensive, off-the-shelf software that isn't built to leverage their unique strengths. Constellation takes a different approach, building software on top of its Atlas graph to enhance your competitive advantages. Info below to get in touch!