- MerlinsNotes
- Posts
- OpenAI co-founder says Cloud+AI is the future
OpenAI co-founder says Cloud+AI is the future
ALSO: Spotlight on building an investment engine using ML
Welcome to the first edition of MerlinsNotes! Hope you’ve had a nice summer and have some great Labor Day plans ahead 😎
I’m starting this series because I’m curious about the various ways in which funds and firms are using AI.
But I’m also tired of all the hype, so I wanted to cut through the noise because I believe it’s important to stay on top of what’s happening with the tech and learn how to master it for your benefit.
My goal is for MerlinsNotes to always be a value-add, so please email me with any feedback along the way!
Here’s what’s on the desk this week:
OpenAI co-founder says Cloud+AI is the future, AI coding assistants are all the rage, and Amazon shocked us all with its GenAI advancements
Building an investment engine for a top PE firm using ML
The ‘best’ way to code with AI
Let’s get into it.
FIRST PASS
Karpathy says Cloud+AI enables scrappy entrepreneurs to run multiple high-value companies leveraging AWS, GPT-4, and Stripe (X)
OpenAI co-founder Andrej Karpathy recently sparked a discussion on X about how Cloud+AI is revolutionizing entrepreneurship.
Commenting on a Lex Fridman interview with popular ‘indie’ entrepreneur Pieter Levels, Karpathy suggested that tools like AWS, GPT-4, and Stripe could enable scrappy individuals to build and run multiple successful companies that could reach billion-dollar valuations.
This trend could make company creation easier and more attractive compared with traditional employment.
Merlin’s Notes: While we don’t think there will be many (if any) ‘billion-dollar, one-person’ companies, it’s worth paying attention to the broader theme here of how Cloud & AI can lower barriers to entry and allow nimble, AI-powered teams to quickly enter markets and capture market share from established players.
Legacy portcos may struggle to match the speed and efficiency of AI-enabled teams, potentially losing ground faster.
To mitigate these risks, funds should focus on strengthening portfolio companies' unique advantages, accelerating AI and cloud adoption, and exploring partnerships with AI tech players to enhance existing offerings and defend market positions.
Amazon’s GenAI tool, Q, saved it an estimated 4,500 developer-years of work (X)
Amazon CEO Andy Jassy recently shared on X how their GenAI assistant, Amazon Q, transformed their internal software development.
The tool dramatically reduced Java upgrade times from 50 developer-days to just hours, saving an estimated 4,500 developer-years of work (we also did a double-take when we read this).
Within six months, over 50% of Amazon's production Java systems were upgraded, with 79% of auto-generated code reviews shipped without changes.
Beyond time savings, the upgrades enhanced security and reduced infrastructure costs, providing an estimated $260M in annualized efficiency gains.
Merlin’s Notes: This is huge and has significant implications for especially tech-heavy fund strategies. AI-powered coding assistants have pulled in ~$1B in funding YoY, making a strong case for the ‘killer app’ title.
And it makes sense. Software engineers are expensive and many companies dream of a future where an AI agent could write and manage all the code they’d ever need.
While that future may never exist (and the jury is out on whether we’d even want such a future), we think PE would be wise to consider prioritizing investments in GenAI tools of their own that can have similarly outsized impacts.
JARGON BUSTER
Retrieval-Augmented Generation (RAG): RAG is a method that enhances AI responses by combining the AI's built-in knowledge with real-time information retrieval.
It’s like giving an AI assistant a personalized library to reference before answering questions.
Here's how it works:
1. Retrieval: The system searches through a database of relevant information.
2. Augmentation: It adds this retrieved information to the original query.
3. Generation: The AI then uses this combined information to create a response.
Think of RAG as a smart research assistant. Instead of relying solely on its pre-trained knowledge, it actively looks up specific, up-to-date information to provide more accurate and relevant answers.
For PE professionals, RAG could be used to enhance due diligence processes, providing AI-powered insights based on the latest market data, company financials, and industry trends.
USE CASE SPOTLIGHT
Building a proprietary investment engine for a top PE firm using public data
Source: Tribe AI
A leading PE firm partnered with Tribe AI to develop a Machine Learning (ML)-driven toolkit for gaining unique insights into a specific vertical using only publicly available data.
The goal? To enhance investment decisions and differentiate themselves in a competitive market.
In just six weeks, Tribe's team of specialists created an MVP that surpassed expectations. They developed a custom ML model that ingested hundreds of predictive factors from public data sources, including census data, social demographics, and infrastructure changes.
The result was a heat map of the US, color-coded at the county level, providing a visual guide for investment attractiveness.
The impact was immediate and significant:
Faster, more confident investment evaluations
Enhanced value proposition to potential portfolio companies
More favorable deal terms in some cases
As the PE firm's SVP of Data noted, "Especially in this market, if you're going to win, then you have to win on speed and price. And the only way you're going to get there is if you have the benefit of the conviction that a tool like this can give you."
The success of this project has led the PE firm to plan expansions into other verticals and incorporate more data sources.
It's a clear demonstration of how data science and AI can be the "last mile of innovation" in PE investing, providing a true competitive edge in sourcing, evaluating, and executing deals.
TOOL OF THE WEEK
Cursor
Source: Cursor
A tweet about building a financial dashboard in <5 mins using Cursor and voice went viral earlier this week and we can see why.
Cursor (built by Anysphere) is a code editor that bills itself as ‘the best way to code with AI’. The company initially raised $8M from OpenAI in 2023 and recently raised $60M from top VCs.
The video in the tweet is a pretty remarkable showcase and provides a glimpse into the future of software engineering.
While the tool is built for developers and those with at least some level of familiarity with coding, one can imagine how anyone with enough desire could use Cursor + Claude/ChatGPT to make something cool.
Have a question about AI? Or need help with a specific scenario?
We’re introducing Ask Merlin to answer all your burning questions!
We’ll read every single question and use it to inform future editions. Submit your question by clicking on the button below.
Until next time! I’d welcome your feedback on today’s edition.
Hit reply to this email or shoot me a note at [email protected]
Thoughts on today's email?No hard feelings either way! |
May the tailwinds be ever in your favor,
— James
P.S. Did someone forward this email to you? If so, you can subscribe here!