The Amity Codex

Axiom 001 · Wrong seat? Move today.

If you feel that you're in the wrong team doing the wrong stuff, don't wait. Don't sit around and wait one week, two weeks, or three weeks. That's a golden rule.

— Nicolas Waern · May 4 · day one

Axiom 005 · Know what to do? Go. Don't know? Bother me now.

If you know what to do, then you don't need to bother me. But if you don't know what to do, then definitely you should bother me instantaneously — or anyone else — until you get what you need.

— Nicolas Waern · May 27

Axiom 006 · Bring options and a default — never open questions.

Don't ask me stuff without options. Come up with a plan — 'Hey Nicolas, we ran into this problem. Either we redo the schema, or we start from scratch, or option three or four. If you don't have anything to say, we're going to go with D.'

— Nicolas Waern · Jun 5 · reprised twice ("three plans, three scenarios")

Axiom 009 · Write your own definition of done.

Create a definition of done. You just tell me what you think you should do, what your definition of done is. I'm not gonna look at it to say yes or no. This is your plan. You go after it, and then you deliver.

— Nicolas Waern · May 27

Axiom 003 · Show up. Early ones get preferential treatment.

The ones that are early are going to get, honestly, preferential treatment, because everyone that is not here is wasting my time and your time.

— Nicolas Waern · May 4

Axiom 011 · Shipping means first users — even one shitty feature.

Shipping means getting first users. Even if it's shitty, even if it's one feature — you have shipped something.

— Nicolas Waern · May 19

Axiom 012 · Don't be like me: 50 features, nothing shipped.

This is my biggest problem, always has been: I'm over here with 50 different features and nothing gets shipped. It gets overly complicated, and the world catches up all the time. Don't be like me when it comes to that.

— Nicolas Waern · May 19 · self-critique as curriculum

Axiom 015 · Deliver something by end of day. It can be as shitty as hell.

Deliver something end of day. It can be as shitty as hell. Just deliver something for the customer to try out. Then you move with their feedback.

— Nicolas Waern · Jun 11

Axiom 032 · Fast beats perfect because feedback compounds.

The whole purpose of Agile is that you ship something really fast with just one feature or two, so the user can say what they want next — instead of you building all the infinite possibilities and then the user doesn't want any of them. I know this because I've tried.

— Nicolas Waern · Jun 26

Axiom 033 · Waiting is a banned word.

Waiting. Waiting. I don't like the word waiting. 'One or two days'? That's a horrible answer. Okay — blame me then. Let's take a meeting right now. In 10 minutes.

— Nicolas Waern · Jul 1

Axiom 035 · Never wait on sign-offs. Best-case assumptions, start anyway.

It's not acceptable to just sit around and wait for stuff. You go with the best-case assumptions and you just get started anyway. You can't wait. It's insanity.

— Nicolas Waern · Jun 11

Axiom 038 · The 5-centimeter rule.

Imagine that right now we're all in the same room. Your teammate is standing 5 cm from your face. You can ask him. Instead, you turn around and look for a cloned version — one month out of date. We're 5 cm away from everyone on the planet. That's what no one is really getting.

— Nicolas Waern · Jul 2

Axiom 037 · Ask forgiveness, not permission.

It's better to ask forgiveness than ask permission. Remember that, please.

— Nicolas Waern · Jul 2

Axiom 044 · AI for format. Zero AI for content that must be you.

Formatting wise, AI of course — make it easy to read. But content wise, zero AI. I need this to come from you.

— Nicolas Waern · Jun 22 · "It was just putting makeup on bullet points."

Axiom 045 · First-prompt exports are disrespectful.

You're in a golden era where you can invoke the spirit of Steve Jobs, Nikola Tesla, Einstein — and instead you say 'hey, tell me something about taxonomies.' It's so disrespectful to everyone in history that hasn't had access to these tools.

— Nicolas Waern · May 15

Axiom 055 · Leave with a team of agents. Non-negotiable.

When you leave this internship, you have to have a team of AI agents. It's non-negotiable, I think, for your future.

— Nicolas Waern · Jun 2 · repeated across three meetings

Axiom 057 · You are supervisors of agents now.

Have an agent. That's your job. You're supervisors of the agents that you create. He's not developing anymore, he's not coding anymore — he's creating loops. That's exactly what we need to do.

— Nicolas Waern · Jun 2

Axiom 043 · Challenge the plan — mine, Claude's, anyone's.

You shouldn't take everything that I said or what Claude had done at face value. Don't take anything that I say or Claude says for granted. Challenge the plan as much as you can.

— Nicolas Waern · May 27 → Jun 5

Axiom 066 · 8% is AI. 92% is scaffolding.

The LLMs, the AI agentic nature — that's 8% of their entire stack. 92% is just pure automation. It's the automation part, the scaffolding and the architecture that is the most important part.

— Nicolas Waern · May 4 · demanded thumbs-up that everyone understood

Axiom 067 · High-stakes means deterministic.

In every high-stakes, high-value industry, everything has to be deterministic. If you ask something 1,000 times, you need to get the same answer a thousand times. If you ask an LLM something 10 times, you're going to get 10 different answers.

— Nicolas Waern · May 4 · "which car would you want to be in?"

Axiom 069 · Robust. Useful. Attractive. Useful wins.

Robust means modular — anything has to be able to be replaced. Useful means it actually does the job — that's what decides if it's good or bad. Attractive means: can someone else understand it? Can AI work on this? Useful is the most important thing.

— Nicolas Waern · Jun 2 · his Vitruvian triad, taught twice

Axiom 071 · One-shot to 90% is a trap.

If you don't understand the architecture end to end, AI will try to hallucinate the hell out of you. You can do a one-shot thing and get to 90%, you think — but then it quickly falls apart when you're going into production.

— Nicolas Waern · May 4

Axiom 082 · Reality is the scaffolding for AI. It already exists.

What is the foundation for AI that we all should be using, that has been there for quite some time? Reality. It's the real world. The scaffolding for AI already exists.

— Nicolas Waern · May 19 · "the most important thing I've ever said thus far"

Axiom 086 · Make reality queryable. Trim 40 years of system fat.

It's making reality queryable and understandable. We're just removing the fat of systems, of Excel sheets, of all this stupidity that we created over the last 30 to 40 years — and going back to reality. For that answer to take 1 minute instead of 2 weeks.

— Nicolas Waern · Jun 3

Axiom 087 · First principles: go to where reality is.

You don't assume that whatever the professors, your teammates, co-workers are doing is correct. You go back and say: where did this start? Why are we doing this? Then you can do whatever you want. If you understand what I'm saying, you can change the world.

— Nicolas Waern · Jun 11

Axiom 090 · Volvo forgot the truck.

At Volvo they had about 100 data scientists all responsible for different parts of the truck. The problem was that they never understood that they were working with a truck.

— Nicolas Waern · May 13

Axiom 098 · The repo is 70% of reality. Capture the other 30%.

70% is what you can capture in the repos, the signals. But 30% is everything else: coordination, orchestration, behavior in meetings. Capturing reality is what it's all about — that most platforms fail to do. And that's why the world is in chaos.

— Nicolas Waern · Jul 3

Axiom 101 · One great page beats ten bad ones.

There's nothing harder than to create a one-pager that is great. It's much easier to create a 10-pager that is shit. You read it, you distill it to the most important thing, and then you reason about it. Then you're learning.

— Nicolas Waern · May 15

Axiom 103 · Less slides, more demo.

Less slides and more demo. Everyone needs to start with a demo. The presentation can't just be text — otherwise it's super boring.

— Nicolas Waern · Jun 18

Axiom 104 · Speak up or planes fall out of the sky.

The stewardesses could see that they were heading into a mountain, but they didn't say anything to the pilots. It's an inability to stand up for what you think is completely wrong — because of culture, hierarchy. That was why the planes dropped from the skies.

— Nicolas Waern · Jun 11 · Korean Air + Challenger O-ring

Axiom 108 · Sell yourself like there's no one like you.

You're at the interview: 'You're going to meet 100 other people, but they're nothing like me. I'm going to tell you why.' That's how I want you to start every interview for the rest of your lives.

— Nicolas Waern · Jun 22

Axiom 116 · Never, never figure it out alone.

You should never, never, ever figure it out yourself. Okay? And I'm saying this to everyone.

— Nicolas Waern · May 19

Axiom 117 · Trees in a forest are never alone.

Trees in a forest, they're never alone. They're always, always connected. They're always communicating and working with each other at all times. So please do that.

— Nicolas Waern · May 7

Axiom 119 · Compete to be first — to help.

By all means, be the first one, and compete: 'Okay, I'm the first one. Does anyone need any help?' You need to stand up in front of class and say: this is what I've done. Does anyone need help?

— Nicolas Waern · May 19

Axiom 121 · Nothing person-dependent. Nothing.

How is that possible that the PowerPoint is tied to one person? Why is this not in a shared area that anyone in the team could present? AI agents, testing, all this kind of stuff — it can't be person and people dependent.

— Nicolas Waern · Jun 16

Axiom 128 · Security: left, right, front and center.

If you don't have a complete interest in security first, for anything and everything that you do, you're not going to get hired anywhere. It doesn't matter if it's front end. It has to be left, right, front and center of everyone's presentation.

— Nicolas Waern · May 11 · repeated across three meetings

Axiom 129 · Test it before you show me. For God's sake.

For God's sake, try it out before you send something to me. And if you're stuck, you tell me you're stuck. How you do it, I don't really care.

— Nicolas Waern · May 27

Axiom 130 · Installed is not working.

I don't want to hear if you installed it. It's like saying you've got a car in my garage, but I can't start it. Saying it's working but obviously it's not — another pet peeve of mine.

— Nicolas Waern · May 19

Axiom 140 · Harden before you extend.

Instead of adding more, make sure that the features that do exist actually work. I want this to be useful and used as soon as possible.

— Nicolas Waern · Jun 26

Axiom 142 · Be obsessed with money — it's fuel for the mission.

You need to be obsessed with money, otherwise you're not going to get anywhere. Seriously. Steve Irwin doesn't care where the money is coming from — he's going to use it to protect whatever he wants to protect.

— Nicolas Waern · May 7 · after 2 of 30 hands went up for "who wants to make shitloads of money"

Axiom 146 · Your reasoning is the only unique thing about you.

The only thing that is unique about you is you and your reasoning capabilities. It doesn't have to be right, honestly. You just have to do it.

— Nicolas Waern · May 15

Axiom 147 · Be more human — AI eats the world.

AI will eat the world. It is eating the world. And what you need to do, as far as I'm concerned, is to be more human.

— Nicolas Waern · Jun 22

Axiom 149 · One life. Get where you want to be.

There's only one life, right? So you'd better figure it out — get to where YOU want to be. Not what your mom or dad or professor or me wants you to be.

— Nicolas Waern · Jun 11

Axiom 159 · The future is unevenly distributed. You're the distributors.

Not even 0.1% use AI. Everyone here is stuck with either analog or Excel — 99% of the market is still stuck in the past. It's our job to bring them into the future.

— Nicolas Waern · Jul 3