Eric Schwartz

Passionate about cycling, running, coffee, and b2b saas. Building AI SRE agents for enterprise @traversal_ai. Prev @togethercompute, @servicenow, @harvardhbs

[Eric Schwartz @eschwartz30]

@anaganath @kleinerperkins @sailresearchco Congrats! I am increasingly convinced that long-running agents are the next frontier and will be a huge unlock for enterprise adoption of agents - excited to follow along and see what Sail builds :)

[Eric Schwartz @eschwartz30]

@trq212 Curious what the benefit of a channel is vs. DM?

[Eric Schwartz @eschwartz30]

@benjitaylor Love the product release video - very cool so see the team emphasizing these UI details in this way!

[Eric Schwartz @eschwartz30]

"The most common error of a smart engineer is to optimize the thing that should not exist."

[Startup Archive @StartupArchive_]

Elon Musk explains his 5-step algorithm for running companies

“First, make your requirements less dumb. Your requirements are definitely dumb… It’s particularly dangerous if a smart person gave you the requirements because you might not question them enough.”

In this interview at Starbase, Elon elaborates on his methodology for shipping everything from electric cars to rockets.

Here’s his “algorithm” quoted in full from the Walter Isaacson biography:

Question every requirement. Each should come with the name of the person who made it. You should never accept that a requirement came from a department, such as from "the legal department" or "the safety department." You need to know the name of the real person who made that requirement. Then you should question it, no matter how smart that person is. Requirements from smart people are the most dangerous, because people are less likely to question them. Always do so, even if the requirement came from me. Then make the requirements less dumb.
Delete any part or process you can. You may have to add them back later. In fact, if you do not end up adding back at least 10% of them, then you didn't delete enough.
Simplify and optimize. This should come after step two. A common mistake is to simplify and optimize a part or a process that should not exist.
Accelerate cycle time. Every process can be speeded up. But only do this after you have followed the first three steps. In the Tesla factory, I mistakenly spent a lot of time accelerating processes that I later realized should have been deleted.
Automate. That comes last. The big mistake in Nevada and at Fremont was that I began by trying to automate every step. We should have waited until all the requirements had been questioned, parts and processes deleted, and the bugs were shaken out.

Elon shares a costly example of doing this process in reverse on the Tesla Model 3 production line and optimizing a part that didn’t even need to exist.

“It’s possibly the most common error of a smart engineer to optimize a thing that should not exist. Everyone’s been trained in high school and college that you answer the question — convergent logic. You can’t tell the professor your question is dumb or you’ll get a bad grade. You have to answer the question. So everyone, without knowing, basically has this mental straight jacket on and they’ll work on optimizing the thing that should simply not exist.”

Source: @Erdayastronaut (Aug 2021)

[Eric Schwartz @eschwartz30]

@StefanStok One potentially out there idea - I had a few Claude sessions where I didn't get the outcome I was hoping for.

Could be cool if Codex could figure this out and either address them, or suggest improvements in general to my ongoing projects.

[Eric Schwartz @eschwartz30]

Testing out the Codex app - very impressed with this onboarding experience. 30 seconds later, everything from Claude Code has been ported over.

[Eric Schwartz @eschwartz30]

After a few hours of playing around with Codex, decided to pull the trigger and cancel Claude for now.

Competition is a beautiful thing - exciting to see OpenAI and Anthropic continue to push each other.

[Eric Schwartz @eschwartz30]

Testing out the Codex app - very impressed with this onboarding experience. 30 seconds later, everything from Claude Code has been ported over.

[Eric Schwartz @eschwartz30]

Big fan of @linear - the product and company that @karrisaarinen and folks like @thenanyu have built is very impressive, and this take makes a ton of sense.

I wonder, though, whether grinding is what is needed in the super early days when you're still hunting for PMF. Sometimes I feel like you just need more shots on goal to hit escape velocity.

[Karri Saarinen @karrisaarinen]

The fallacy of this is that more creates more. More hours, more hiring, more something.

And it is true in a sense. If you put in more work, more work will happen. But I think for most startups, the leverage is really in how differently you approach the problem, how well you cultivate your team, and the strategy.

Any large company can outspend you on hours. They have thousands or tens of thousands more people, spending more hours. If hours worked were the metric, every large company and government organization would always win and do the best work. More hours, better output.

This thinking is often representative of younger founders, where the startup becomes their identity and life. They have a hard time doing anything else, and cannot understand that your work is not the person that is you. But activities outside of work can grow you as a person too and make you do better work.

I’ve never worked this way. As a designer, I always saw the need to take a step back, to take a break. At times, I might work 12 hours or 16 hours, or whatever amount was needed, but it wasn’t the norm. You just can't grind design, you need inspiration. But taking that step away from the work, would give me more perspective, inspiration and I could approach the problem differently or I could just see the solution.

Grinding is never good for any creative problem, and startups or creating new products are often mostly about creative problem solving. Grinding works ok for email jobs, or where you just executing on very clear playbook.

With Linear, we’ve never worked this way. We work reasonable hours, 5 days a week. All of us founders have families. Many of our employees have families. I personally stop every evening, spend time with the family, cook dinner for the family, eat dinner together, and focus on things outside of work. Sometimes I work in the late evenings or weekends, but to me the pride is that I don’t need to. Company should be succesful without it.

My goal is to build a company that is sustainable in the long term, and doesn’t require heroics or personal sacrifices every single day.

There are times when our team is heroic. Launches, incidents, some other work that just needs to be done. They will work late into the night because they know it is the right thing. But we don’t require that every day or every week, and the more this happens, the more I think it is a failure of our company and leadership. The team and the leaders should always keep a reserve to use when something is needed.

Our thinking was also that quality, which we value, doesn’t emerge from working more or stressing people more. It emerges when you create the conditions for it to emerge. Often it is the appreciation, space, time, and how the person feels. A person who is rested will do better work.

I wouldn’t attribute much of our success to working a lot. The success came from having clear thinking, ideas, and focus to do the right things.

I sometimes wish we could move the culture more toward a Zen master.

Real mastery is not exerting the most effort. It is achieving the outcome with the least necessary effort.

[Eric Schwartz @eschwartz30]

Well I'm hoping you can get back to great performance on 1M context with 5.5 soon :)

R.e., plan/goal mode - it might be that I'm used to Claude Code in the terminal and just hitting shift+tab to toggle to plan mode and then back to skip permissions mode. They feel weirdly isolated in a separate menu.

[Eric Schwartz @eschwartz30]

@danshipper Just switched from Claude to Codex today after listening to your podcast with @lennysan - very impressed!

[Eric Schwartz @eschwartz30]

@nunezvice @reach_vb Really loving the in-app browser preview. Claude's functionality is much more limited which is why I was tempted to switch.

Some suggestions:
1. PLEASE add a 1M context window
2. Hiding plan mode and goals separate from permissions seems a bit odd to me

[tobi lutke @tobi]

https://t.co/5MJ7u1VHwf

[Eric Schwartz @eschwartz30]

More than ever, I am extremely skeptical when I hear that something cannot be done because of some "rule" or "best practice" without a clear explanation for why. The only rules that can't be broken are laws of physics. Everything else is just a recommendation or an opinion.

[Benji Taylor @benjitaylor]

We need to stop talking about product design in absolutes… there are no rules. Everything is made up. Do what makes sense and feels right.

[Eric Schwartz @eschwartz30]

@ali_aka_ahmed I've been so curious what these numbers represent! Revenue?

[Eric Schwartz @eschwartz30]

My girlfriend has realized that the best way to get a hold of me when I'm deep into Claude Code / not checking my phone is to email me. I get these emails a few times a week now

[Eric Schwartz @eschwartz30]

Me heading to work on Monday

[Dan Pantelo @danpantelo]

You came to NYC to ‘build’ but you’re skipping office today bc of a travel ban? In the agentic era??

Embarrassing, low agency. It’s time to escape the permanent underclass.

18-24 inches? That’s how deep Claude will be inside my workflows today.

[Eric Schwartz @eschwartz30]

Podcast link: https://t.co/WoGgpUrniN

One of your best episodes yet @lennysan @bcherny!

[Eric Schwartz @eschwartz30]

I've been reflecting a lot on how product / design has changed for AI products vs. traditional SaaS. Can't think of a better primer on this subject than @lennysan's recent interview with @bcherny

This should be required listening for anyone building in tech. Two lessons stood out to me in particular 🧵

[Eric Schwartz @eschwartz30]

2. Latent demand hides in "misuse"

The best way to uncover use cases is to watch how people "misuse" your product. But that requires building something flexible enough to be misused in the first place. If it's too restrictive — or forces people to totally change their workflows — you'll never discover the use cases you never thought of. You have to meet people where they are.

The same principle applies to models. Give LLMs flexibility, observe what they want to do, then make it slightly easier over time — rather than restricting them with too much scaffolding upfront.