Build Brains

Happy Thursday. It’s Kentucky Derby week. Betting on horse racing is one of the best ways to spend a Saturday (if you’re okay with losing a bit of money).

The favorite, Journalism (#8), is probably your best bet. If you want a value bet, Sandman (#17) is your horse. If you are a real nut case a want an exotic bet, I would do an exacta box with the following horses: #8 Journalism / #17 Sandman / #9 Burnham Square / #18 Sovereignty. I like to think I’m good at horse betting, but there is definitely a 75% chance you lose money. You’re welcome.

Proprietary Knowledge

When you start a new job, you are given an overview of all functions within the company. You hear about the goals, overall mission, and what other teams are doing. After this quick presentation, you usually get locked in a silo of information and knowledge that is specific to your own team and you lose touch of what other teams within your company are doing. Most companies like to say they are “cross-functional” but this just doesn’t happen, and it’s a reality of most jobs.

The longer you stay at a company in a specific role, the longer you are building up your internal knowledge and data in your own head. Independently within each team this keeps building up. I am by no means arguing that people should spend time understanding what every team and person in the company is doing, that’s almost as big a waste of time as doing HR training. However, the internal brains that companies build within teams can become disconnected from others and end up swimming in their own lanes.

Getting stuck in your own lane could mean that you are just extremely efficient at doing your job and understand how sh*t works. This is all good, but disconnecting from the rest of the company could open up other vulnerabilities.

Build AI Brains

This is likely my biggest recommendation I can make to startups today. Build an internal brain(s) for your company by utilizing AI. It doesn’t need to combine every team's knowledge into one. Sales, product, and marketing could be combined, while there is a separate one for just marketing and customer success. The goal is to combine multiple data points into one place so you have a true source of knowledge for any specific function. Essentially, this becomes a knowledge network power by AI.

At the end of the day, the entire corporate world and economy is built on data. Qualitative or qualitative, it can all be thrown into spreadsheets, PDFs, and documents that can be used as context for a LLM to use when answering questions. Likewise, you can connect internal systems to a data repository that acts as your context here.

Side bar

Whenever I say context from now on when relating to AI, think “brain”. AI is extremely good at taking data, whether it be PDFs, actual data, or call transcripts, and giving you back incredible outputs. For companies, when I talk about giving an AI context, I am talking about essentially giving the AI a brain of knowledge that it can answer questions from. When you ask the AI a question, it will use the context you gave it instead of pulling an answer out of its ass. 

Back on Track

So if the idea is to connect all of this data into one place, what is the desired outcome? It really depends on your end goal.

Maybe you want more insight into what your customers are saying, so you build a brain that is fully based off of transcripts from client calls.

Maybe you want to understand what the product currently can do and the roadmap, so you build a brain based off of past product documentation and future roadmap planning.

Maybe you want to build better marketing material based on product and sales calls, so you build a brain containing great product reviews and client calls.

All possible, it just depends on your goals and what you want to build. I will say, having built some of these before, it is extremely fun when you first use one after building it. It feels like magic. 

Other Areas

If you couldn’t tell, I am very excited about the application of this technology and idea when it comes to startups. Doesn’t mean that larger companies won’t implement it at some point, but the amount of policies and hands that have to touch a product like that in a big company will cause it to take years to release. 

Nonetheless, this practice of building an internal brain is being practiced across the finance industry as we speak. In fact, just today Rogo AI, an AI purpose built for being a financial analyst, raised a $50m Series B on a $350m valuation. This comes just a few months after their $20m Series A priced at an $80m valuation. 

While Rogo focuses on large investment banks, private equity firms, and hedge funds. There is a gap in the market at the lower end. Family offices, boutique PE shops, and smaller funds don’t need to invest in other software to power their AI practices. They can be built in-house.  

Most of these firms have years of data, investment reports, and internal discussion that can be fed into an AI as context for moving forward. In this way, an internal AI built in these firms can almost become their own financial analysts that helps them evaluate new deals and opportunities based off of previous data that has (hopefully) led to some success. 

There is a rabbit hole I could go down here, but that’s for another time.

Song of the Week