I think AI agentic workflows will drive massive AI progress this year — perhaps even more than the next generation of foundation models. This is an important trend, and I urge everyone who works in AI to pay attention to it. Today, we mostly use LLMs in zero-shot mode, prompting a model to generate final output token by token without revising its work. This is akin to asking someone to compose an essay from start to finish, typing straight through with no backspacing allowed, and expecting a high-quality result. Despite the difficulty, LLMs do amazingly well at this task! With an agentic workflow, however, we can ask the LLM to iterate over a document many times. For example, it might take a sequence of steps such as: - Plan an outline. - Decide what, if any, web searches are needed to gather more information. - Write a first draft. - Read over the first draft to spot unjustified arguments or extraneous information. - Revise the draft taking into account any weaknesses spotted. - And so on. This iterative process is critical for most human writers to write good text. With AI, such an iterative workflow yields much better results than writing in a single pass. Devin’s splashy demo recently received a lot of social media buzz. My team has been closely following the evolution of AI that writes code. We analyzed results from a number of research teams, focusing on an algorithm’s ability to do well on the widely used HumanEval coding benchmark. You can see our findings in the diagram below. GPT-3.5 (zero shot) was 48.1% correct. GPT-4 (zero shot) does better at 67.0%. However, the improvement from GPT-3.5 to GPT-4 is dwarfed by incorporating an iterative agent workflow. Indeed, wrapped in an agent loop, GPT-3.5 achieves up to 95.1%. Open source agent tools and the academic literature on agents are proliferating, making this an exciting time but also a confusing one. To help put this work into perspective, I’d like to share a framework for categorizing design patterns for building agents. My team AI Fund is successfully using these patterns in many applications, and I hope you find them useful. - Reflection: The LLM examines its own work to come up with ways to improve it. - Tool use: The LLM is given tools such as web search, code execution, or any other function to help it gather information, take action, or process data. - Planning: The LLM comes up with, and executes, a multistep plan to achieve a goal (for example, writing an outline for an essay, then doing online research, then writing a draft, and so on). - Multi-agent collaboration: More than one AI agent work together, splitting up tasks and discussing and debating ideas, to come up with better solutions than a single agent would. I’ll elaborate on these design patterns and offer suggested readings for each next week. [Original text: deeplearning.ai/the-batch/is…]

delete social media read a chapter a day make friends with older people / interesting folk ponder by the sea take an online class so you “learn” how to study again eat well and workout find a nice lady

Humanoid robots are coming... Trillion dollar question is whether a few companies will capture the economic profits or this will become similar to the EV industry (low margin, high competition, high capex business with price wars)? bloomberg.com/news/articles/… via @BW

Will US Chips Act $$ be successful in helping build competitive chips fabs in the US? A must listen talk at MIT by Morris Chang - Founder of TSMC - about the structural challenges to onshoring modern chip production. BG2 Ep3. @bgurley @altcap @SecRaimondo @NVIDIAGTC

Important to clarify... If @saylor executes his strategy well it is entirely probable MSTR outperforms #Bitcoin in fiat terms consistently for a very long time. However, there are numerous single points of failure with MSTR stock. There are multiple ways this strategy backfires and the stock underperforms #Bitcoin. Adjusted for risk, #BTC in proper self custody is by far the better thing to own. It is entirely possible to be extremely bullish on @MicroStrategy stock because you think @saylor and the company will execute their strategy well, yet still understand #BTC is the more important asset to own and the better risk-adjusted asset to own... MSTR is the better fiat returns than #BTC on the condition the humans behind MSTR perform well. MSTR is probably the best bet in the world on an entity controlled by humans. However - #Bitcoin's monetary policy isn't controlled by humans at all... @MicroStrategy may be a very good (or the best) investment, but it is not money.

to build a great b2b company you, you need a founder who believes that distribution & sales are equally as important as the product. a great gtm motion starts from the top down–not just in theory, but in action. what it means to prioritize go-to-market: → revenue teams have real input to the product roadmap because they are the voice of the customer → you’re indexing strong engineering talent towards go-to-market → you’re hiring people on go-to-market who are as hungry for the company’s long-term success as the engineers are the focus on sales & distribution has to come from the founder’s dna, otherwise it won’t seep into the culture of the org. this is something that the ramp founders nailed. and that mindset has stuck with me as I’ve been building @unifygtm over the past year. plg is an important strategy. it works. but you have to embrace sales if you have plans to build a generational company. do you agree?

b2b startups: true early adopters do not want a complete solution read that one more time they want a quick, easy way to capture a unique, competitive edge before anyone else in the market can extract the value laggards want complete solutions therefore, overbuilding will repel you from true early adopters AND anyone asking/requiring more features is likely wasting your time (i.e., tire kicking) something not often discussed the spark of insanity from a founder can vicariously be felt by the market that excitement/enthusiasm of specific future value is what early adopters buy into, albeit if it's aligned with a problem they recognize