aelaguiz

Maintaining huge Markdown files sucked, and kept introducing subtle issues as they drifted over time.

Python agent frameworks like crew aren't portable between different harnesses.

So I built Doctrine. I couldn't go back.

It turns writing agentic Markdown into coding, and my codex can now do a great job of building workflows using the doctrine language.

I use it for codex, openclaw, hermes and paperclip agents currently. My workflows have gone from "mostly working" to bullet proof.

It ships with a vs code / cursor syntax highlighting plugin as well.

I hope someone else finds it useful!

I'd love to chat!

If you wanna shoot me a note I'd love to hear any thoughts you have!

Amir@pokerskill.com

I know this is driving me nuts! I've had the fix in the approval queue just waiting on Apple :)

If you wanna shoot me a note I'd love to hear any thoughts you have! Amir@pokerskill.com

This is my area of expertise. I love the experiment.

In general games of imperfect information such as Poker, Diplomacy, etc are much much harder than perfect information games such as Chess.

Multiplayer (3+) poker in particular is interesting because you cannot achieve a nash equilibrium (e.g. it is not zero sum).

That is part of the reason they are a fantastic venue for exploration of the capabilities of LLMs. They also mirror the decision making process of real life. Bezos framed it as "making decisions with about 70% of the information you wish you had."

As it currently stands having built many poker AIs, including what I believe to be the current best in the world, I don't think LLMs are remotely close to being able to do what specialized algorithms can do in this domain.

All of the best poker AI's right now are fundamentally based on counter factual regret minimization. Typically with a layer of real time search on top.

Noam Brown (currently director of research at OpenAI) took the existing CFR strategies which were fundamentally just trying to scale at train time and added on a version of search, allowing it to compute better policies at TEST TIME (e.g. when making decisions). This ultimately beat the pros (Pluribus beat the pros at 6 max in 2018 I believe). It stands as the state of the art, although I believe that some of the deep approaches may eventually topple it.

Not long after Noam joined OpenAI they released the o1-preview "thinking" models, and I can't help but think that he took some of his ideas for test time compute and applied them on top of the base LLM.

It's amazing how much poker AI research is actually influencing the SOTA AI we see today.

I would be surprised if any general purpose model can achieve true human level or super human level results, as the purpose built SOTA poker algorithms at this point play substantially perfect poker.

Background:

- I built my first poker AI when I was in college, made half a million bucks on party poker. It was a pseudo expert system. - Created PokerTableRatings.com and caught cheaters at scale using machine learning on a database of all poker hands in real time - Sold my poker AI company to Zynga in 2011 and was Zynga Poker CTO for 2 years pre/post IPO - Most recently built a tournament version of Pluribus (https://www.science.org/doi/10.1126/science.aay2400). Launching as duolingo for poker at pokerskill.com

Whoever didn't get those contracts must be unhappy. That's how I read this type of story.

Yes we all want the benefits produced by incentive structures that bring out the best of humanity with absolutely no harm or asymmetric benefit.

This junk always gets attention because people don't read history therefore they have no basis for comparison.

The irony of posting this on the literal celebration of upwards mobility and capitalism that is YC is delicious.

The whole article can be summarized by reading the last sentence: "some problems were just not meant to be solved."

That's it. The author doesn't believe the on-demand economy can work. The rest is just lead up.

I disagree with his assessment.

Based on this I can only assume that workiva builds spaceships or life support systems. You know, something where absolute perfection is far more important than having a fulfilling and constructive workplace environment.

I’m sure a lot has changed since 2014 when I last raised in Austin but in my experience it is possible to raise money for consumer businesses in Austin. Before we pivoted to Cratejoy we had raised a significant seed round from almost entirely Austin investors. It’s also possible to get west coast (and really all over the world) investors to invest in you and stay in Austin. We’ve raised $11M, about 80% of it from non-Austin investors.

For me the secret for raising successfully in Austin was getting individuals on board — high net worth individual investors that I established a personal rapport with. With the exception of the Capital Factory all of the angel funds (for me) were operating off of FOMO and social pressure more than fundamentals. The VC scene does contain some bad actors, and I probably wouldn’t spend a lot of time looking for institutional money in Austin. I would get my growth funding west or east coast.

Having raised millions in both SF and in Austin I can say that it is definitely easier in SF. However, the cost of living is so much lower (and I would say quality of life is so much higher) in Austin that I’m glad we’re based here. It’s also relatively easy to recruit from all over the world to Austin, everyone has seen the continuous stream of articles rating Austin as one of the best places in the US to live.

There’s also another great side-effect of being Austin which I didn’t notice until we were already scaling very quickly. There’s a smaller number of startups in Austin that are producing millions in revenue very quickly (as compared to SF) but the enthusiast and talent pool is very large. When we started to hit scale we had way more advocates than we ever expected. Austin wants home-town winners and it throws its weight behind the ones that look promising, but you have to have something to show.

My cofounder and I consider having Cratejoy based in Austin a competitive advantage.

It's like someone from a future in which they can perfectly model the human brain and predict it's reactions wrote an article designed to perfectly stimulate all of the anger centers of my brain.

(I'm referring to the qz article, not pg's essay)

This is a much better version of the original article. Here's solutions to situations that arise when you have access to information that you feel should be acted upon that doesn't have broad organizational support already.

Cratejoy (YC S13) | Austin, TX | Full-time | ONSITE

Cratejoy has created a new ecosystem of subscription businesses. Thousands of first time founders and new business owners have created subscriptions on Cratejoy with hundreds of thousands of subscribers. We've recently launched a shopping portal for consumers to find subscriptions operated by Cratejoy merchants. We have raised about $11M from the best investors in the valley and are driving millions of dollars in revenue.

Cratejoy is literally creating thousands of new jobs and an entire new set of businesses, come help us build something that people truly love.

We're hiring:

- Platform Software Engineers (Python/Flask, postgres, AWS)

- Full stack Software Engineers (Backbone + REST Python backend)

- Front-end Software Engineers (Backbone, Bootstrap, LESS)

- Support Engineers (Python, Javascript)

- Data Analysts (Mixpanel, Postgres, Python)

- Product Managers (Marketplace or SaaS)

We are willing to pay for relocation expenses to Austin.

Apply at: https://jobs.lever.co/cratejoy

Feel free to e-mail me with any questions about working at Cratejoy: amir@cratejoy.com

It seems like many of the commenters here are confusing corporate strategy/tactics with individual mean-ness. I'm not saying its good when a corporation takes action to derive profit at the expense of the public interest or that of one of their competitors - however, it's not nearly the same as an individual being malicious or petty when given power over others (such as a founder routinely has).

That makes total sense.

We did a syndicate on AL that condensed a $500k round into a single LLC. That was nice. The rest are individuals.

We have 20 something investors and on a given month there are ~2 that send more than one e-mail, 5 that send 1 e-mail, and basically nothing from all of the rest. I don't feel a burden, tbh. I like it, when I need help I get it - and I have a few really solid mentors in the group.

The only pain is at series A when I'm going to have to go get all of them to sign the closing docs. That'll be hard.

Great article Bemmu.

I love the professional product photos and the clear "what last month's box" contained photo on your home page. Some of our merchants @ Cratejoy have been experimenting with including a "last month's" box section/photo and have been getting very favorable results.

I'm curious, what is your monthly retention?

Have you done experiments with box contains and how it impacts retention?

We're starting to run optimization experiments where we take the product inventory and try to optimally allocate it amongst all subscribers to maximize retention. So like month 1 subscribers get the exact optimal set of products to make sure they stay subscribed until month 2, etc.

It's awesome that you are doing product surveys. I hope you do them every month. Have you ever considered sharing the results of your product surveys with your suppliers? Frequently that will be a great relationship building thing, maybe even get you better pricing or free product.

Congrats guys!

I love this company. We use them every day and we recommend them to all of our merchants (who in turn love them). Their product is in a small but distinguished group of tools that are so good they set a new standard. I consider circleci amongst that group, for reference.

Congrats guys

A CTR of 0.4% is absolutely abysmal for a newsfeed ad and the fact that the author thinks that it is "quite high for an ad like this" makes me think he has no idea what he's talking about.

Congrats Vlad, Bryant & Sergie. Killer product, great raise.

Congrats Sam. I love your thoughtful writing style and think you're the natural choice for the job.

The thing that makes me happy is sitting in a room with a few people I really like and building something I'm interested in. We already all live in nice houses and have the computers we want. We can eat wherever we'd like. If I want to take a vacation I can afford it.

How will having more money make it so I can better sit in a room with a few people I like and build stuff?

I mean it's not like being wealthier will make my Netflix have cooler stuff in it.

Totally true. I'd add though that there definitely is a level of initial respect you get as a YC company. It doesn't help close the deal, but it will generally keep you out of associate land.

I sometimes wonder if people like this wake up one day and realize they tossed one of the most important opportunities of their lives because they are stubborn. Probably not.

Just under 2 years, it was a nightmare. I gained weight and became depressed. I left and will never sell a company like that again. If there's an earn out it'll be 1 year, otherwise I won't do it.

Yeah that doesn't matter to me.

Maybe others experience wealth differently but it made substantially no difference to my day to day happiness (once the newness of the money had worn off) - but working on a project I feel strongly about makes a strong and lasting difference.

Yeah I didn't say people past their 20s don't aim high. I said your 20s is a great time to aim high.

Love it!