six*

splitting my chief of staff agent across five undifferentiated providers so i can't be killed

everyone chill out. the computers yearn to help us

Sharing an interesting recent conversation on AI's impact on the economy. AI has been compared to various historical precedents: electricity, industrial revolution, etc., I think the strongest analogy is that of AI as a new computing paradigm (Software 2.0) because both are fundamentally about the automation of digital information processing. If you were to forecast the impact of computing on the job market in ~1980s, the most predictive feature of a task/job you'd look at is to what extent the algorithm of it is fixed, i.e. are you just mechanically transforming information according to rote, easy to specify rules (e.g. typing, bookkeeping, human calculators, etc.)? Back then, this was the class of programs that the computing capability of that era allowed us to write (by hand, manually). With AI now, we are able to write new programs that we could never hope to write by hand before. We do it by specifying objectives (e.g. classification accuracy, reward functions), and we search the program space via gradient descent to find neural networks that work well against that objective. This is my Software 2.0 blog post from a while ago. In this new programming paradigm then, the new most predictive feature to look at is verifiability. If a task/job is verifiable, then it is optimizable directly or via reinforcement learning, and a neural net can be trained to work extremely well. It's about to what extent an AI can "practice" something. The environment has to be resettable (you can start a new attempt), efficient (a lot attempts can be made), and rewardable (there is some automated process to reward any specific attempt that was made). The more a task/job is verifiable, the more amenable it is to automation in the new programming paradigm. If it is not verifiable, it has to fall out from neural net magic of generalization fingers crossed, or via weaker means like imitation. This is what's driving the "jagged" frontier of progress in LLMs. Tasks that are verifiable progress rapidly, including possibly beyond the ability of top experts (e.g. math, code, amount of time spent watching videos, anything that looks like puzzles with correct answers), while many others lag by comparison (creative, strategic, tasks that combine real-world knowledge, state, context and common sense). Software 1.0 easily automates what you can specify. Software 2.0 easily automates what you can verify.

IDEA ABSORPTION Idea generation is one of my biggest strengths as a problem solver but the inability to watch when / where i was doing it, was a big weakness early in my career as an executive / leader. I didn’t even realize I was inundating my org with ideas, until a peer Eng leader came to me and said “Gokul you’ve got to stop generating idea confetti. The org needs focused execution, not more ideas.” I’ve never forgotten that line from nearly two decades ago. I made two changes: (a) ideating in very small, mostly 1:1,settings with senior leaders, to flesh out a concept (I do this with AI today!), vs in groups, unless the goal of the group meeting was to ideate, and (b) trying to be Socratic in phrasing: “should we consider X” and being clear that X is a suggestion, not a directive, and it’s the team’s job to decide if X makes sense vs all the other things on their plate. Learnt this the hard way after being presented work output by someone who had worked on a project for a week based on misinterpreting a throwaway comment from me at a meeting. It was cathartic to see that Bezos had the same issue early on in his career. From a recent interview: “Jeff Wilke came to me one day he worked for Amazon for a quarter of a century but this is when he probably knew me only for a year and he said “Jeff you have enough ideas to destroy Amazon.” And this was such a shocking idea for me Jeff said you have enough ideas per minute per day per week to destroy Amazon. I was like what do you what do you mean? He’s like you have to release the work at the right rate that the organization can accept it and he was a manufacturing expert and so you know his view of the world was every time I released an idea I was creating a backlog a queue work in process and because it was just stacking up it was adding no value and in fact it was creating distraction and so he said ‘Look you have to figure out when to release these new ideas at a rate that the organization can accept them and this was I mean this sounds so obvious but it was not obvious at the time to me and this was a profound insight for me and so I started prioritizing the ideas better keeping lists of them keeping them to myself until the organization was ready for the ideas and then I also started figuring out how can I build an organization that can be ready for more ideas that’s about having the right senior team and the right leadership and getting those people the executive bandwidth so they could do more ideas per unit time and so and and that is what we built we built a company that’s very good at inventing and doing more than one thing at a time and you do want to build as the company gets bigger you do want to be able to do more than one thing at a time but that idea of releasing the work was very profound for me and it made it made us operationally more effective while still being inventive and do you think you’re a better inventor.”

This problem area has bounced around in the back of my mind as an obviously great opportunity -- so excited to see @omarjpeg , Nathan, and team executing on it so well and so quickly!

It's true, claude is better than writing in OpenAI's o1 model. However, o1 can do something claude cannot: process 3-4 hours of transcripts and create detailed worksheets, summaries, and FAQs that cite what the speaker has said word for word, and recall every question answer pair.

I’m hiring Product Designers for @airtable AI! We’re creating tools to make AI actually work for work, helping teams easily connect data, generate insights, and automate workflows.

the beauty with enterprise sales is you just need ~5 customers as references to unlock the next 20 referenceability is how you unlock scale in the enterprise those first few enterprise buyers are worth 100X what they're paying ...