Build Code Practice

If you build things for other people and you want a practical framework for making your values visible in every decision you make, this is for you. If you've ever drifted toward a shortcut that compromised the people you were building for and needed language for why that felt wrong, this is also for you.

What is a build code?

Most people who build things have values. Not everyone makes those values explicit.

A build code is the practice of doing exactly that: translating your ethics and your design instincts into a document that governs every decision you make, from database schema to system prompt to the words on a consent form. It is not a privacy policy. It is not a set of design principles. Those things exist to protect the provider or guide the user. A build code exists to keep the builder in integrity.

Think of it as a compass. It doesn't tell you every step. It keeps you oriented when the messy middle of development gets hard, when commercial pressure, time constraints, or technical limitations tempt you toward shortcuts that compromise the people you're building for. Without it, values stay implicit. And implicit values in tech tend to default toward what is easiest, most profitable, or most familiar, which is rarely centred on the people most often excluded.

This is not a framework for everyone. If you're building tools to maximise engagement or push a particular narrative, this is not what you're looking for. But if you want to call yourself an ethical tech developer, if you want your technology to shift power, not just money, then having a code you live and build by is not optional. It is the thing that keeps you from drifting toward harm without noticing.

What I know from eighteen months of building is that a build code is not something you write and then execute. Dated study group evidence of each shift is in Appendix C3. It is something you discover under pressure. Three moments across four builds shifted mine from aspiration into architecture:

Each of these produced a clause in the build code. None of them came from planning. They came from breaking something I cared about and having to name what I refused to break again.

The Master Chief: who is building this?

My name in every AI tool I use is Master Chief. I love the Halo show and I love what Master Chief represents, a principled operator who holds a code even when the mission gets complicated. It is my alter ego as a builder.

Developer_Dossier.md

Master Chief

Systems-thinking Storyteller

CliftonStrengths Profile

Ideation • Learner • Futuristic • Activator • Focus

Design Philosophy

"Every line of code carries an opinion."

Non-Negotiables

• const No extraction without reciprocity
• const Absolute data sovereignty
• const Integrity is hard-coded
• const No automating human presence
• const No cultural performance

Still Learning

• → Somatic sustainability (avoiding the Red Zone)
• → Cultural humility in Māori spaces
• → Staying curious longer before jumping to solutions

My strengths profile (CliftonStrengths) is: Ideation, Learner, Futuristic, Activator, Focus. I am a systems-thinking storyteller. I think in problems I want to solve and stories I want to tell. My operating system is grounded in Kaupapa Māori te reo Māori Core Value Kaupapa Māori A Māori-centred approach to research — research done by, for, and with Māori communities. Research Context "It is not a method bolted onto a Western framework. It is the framework." and Pacific values, specifically manaakitanga te reo Māori Core Value Manaakitanga Care as obligation, not gesture. Research Context [1, 4] ↗ "In this research, manaakitanga lives in the AI's pacing, its validation, its insistence on checking your nervous system before asking about your relationship. It is care in the code." (care), whanaungatanga te reo Māori Whanaungatanga Relationship, kinship, the bonds that make people belong to each other. Research Context [1, 4] ↗ "In this research, it means prioritising the relational bond over data extraction. Trust before content. Connection before questions." (relationship), and kaitiakitanga te reo Māori Core Value Kaitiakitanga Guardianship — stewardship that extends across generations. Research Context "In this research, it means protecting data and knowledge not just for the people in the room today, but for their mokopuna (grandchildren)." (guardianship). These aren't values I borrowed for this project. They're the reason the seven-dimension framework below looks the way it does. Manaakitanga te reo Māori Core Value Manaakitanga Care as obligation, not gesture. Research Context [1, 4] ↗ "In this research, manaakitanga lives in the AI's pacing, its validation, its insistence on checking your nervous system before asking about your relationship. It is care in the code." shaped how I defined purpose, whanaungatanga te reo Māori Whanaungatanga Relationship, kinship, the bonds that make people belong to each other. Research Context [1, 4] ↗ "In this research, it means prioritising the relational bond over data extraction. Trust before content. Connection before questions." shaped what I count as safety, kaitiakitanga te reo Māori Core Value Kaitiakitanga Guardianship — stewardship that extends across generations. Research Context "In this research, it means protecting data and knowledge not just for the people in the room today, but for their mokopuna (grandchildren)." gave me the language for data sovereignty.

I believe every line of code carries an opinion. I refuse to build "tech for tech's sake." I define myself partly by what I refuse to do: I will not build persuasive tech designed for addiction. I will not automate human connection where human presence is required. I will not compromise data sovereignty for convenience.

The moment I am most proud of as an expression of my build code was the Incognito Mode feature in the Leadership AI Coach. I built it myself, not because anyone told me to, but because I had just launched the Ray pilot and experienced the full ethical weight of what I was holding. Users were coming in vulnerable. They were crying. They were disclosing things that had nowhere else to go. And I understood, viscerally, that they needed to be able to engage without any trace being kept. Not a privacy policy that said it was safe, but a physical mechanism that made it structurally impossible for a record to exist. A frontend toggle that blocked the entire logging API. Zero data generated. Purely private. That was mana motuhake te reo Māori Core Value Mana Motuhake Absolute sovereignty — over your data, your story, your identity. Research Context [1, 3, 4] ↗ "In this research, it means the person who generated the data owns it. Every decision about what gets stored, who sees it, and whether a record exists at all traces back to mana motuhake." made real in code. For the full technical implementation, see Building Safe Conversational AI.

The build code framework

Across all four builds, seven dimensions proved essential. The values-to-architecture map showing how these dimensions played out in practice across builds is in Appendix A4. Each one is both a question and a commitment:

Build codes per implementation

Each build had its own specific code, shaped by its vulnerability level and the communities it served.

On "discomfort is data": When I demonstrated the AI mispronouncing te reo at the start of the wānanga te reo Māori Core Value Wānanga A gathering for deep learning and knowledge sharing. Research Context "In this research, the Culture Meets AI wānanga was a 90-minute session where participants explored together whether AI belongs in cultural spaces." , I did it deliberately. I said, "It's giving me the ick" and everyone laughed, and then we got it out of the way. I didn't want people sitting with that discomfort for the whole session. Naming it early, letting the AI acknowledge its own limitation, and moving on was a facilitation decision driven by the build code: discomfort is a signal to go deeper, not to avoid.

On relational vs. procedural safety in Ray: Relational safety was that I knew everyone in the study, or they were participating with a trusted person who knew they were in the pilot. I offered a debrief after every session. Procedural safety is the guidelines in the background, the flags, the webhooks, the crisis escalation. But relational safety came first, because you can't bolt on trust.

The "NO" clause: refusal as ethical design

The most deliberate design decisions I made were the ones about what I refused to build.

An explicit NO clause lets you pre-commit to refusal before the pressure arrives, before the commercial ask, the shortcut, the "just this once." It protects the mana of the user and the integrity of the builder in the same move. Without it, you can drift toward harm without even noticing. Look at the Groks of the world: vulnerable users are not at the centre. Eyeballs and engagement are. In some cases it's not even about money; it's about pushing a particular rhetoric. A build code doesn't prevent that in others, but it makes it impossible for you to end up there without knowing you made the choice.

My NO clause evolved across four builds. The line about "no automated empathy that claims a heart it doesn't have" came directly from participant feedback, a pattern of comments across transcripts that made me realise the AI was at risk of performing care rather than facilitating it. AI agents do not have heart. They simulate. The intention behind that simulation is everything. If you're doing it to build false intimacy and extract value from someone, that is harm. If you're doing it to create a safe space someone has chosen to enter, that is a design decision. The line between them is the NO clause.

One thing I learned specifically from Ray: early on, before the pilot, a friend told me Ray's voice was "so dreamy" and she'd "found her new boyfriend." That was a hard stop. The NO clause for Ray had to include an explicit prohibition on romantic framing, on relationship-forming persona, on anything that could encourage emotional dependency. I updated the system prompt, added guardrails, and reframed Ray clearly as the "wise mate on the back porch."

W-10 described what governance needs to look like for decisions this consequential: "My first and foremost, most important thing for them would be to, once this is created, or even before, there should be a pre, a current, and a post engagement. This needs to be taken to iwi members, kaumātua, kuia, leaders for their input. And why I say this is because they are able to identify what tapu te reo Māori / Framework Terminology Core Value Tapu & Noa Tapu is the state of being sacred, restricted, under spiritual protection. Noa is the state of being ordinary, accessible. Research Context [2, 4, 8] ↗ "The central paradox of this research sits between them: AI risks making tapu things noa. But for people cut off from their culture, noa may be the only door available. That paradox is not resolved. It is held." is and what is not, and what shouldn't be touched and what can be." That is the ohu/ope model described by a participant, not a researcher. The NO clause protects the builder's integrity. The governance structure protects the community's mana. Both are needed.

The DreamStorm / Trurivu Kaupapa Charters

DreamStorm started as my digital sleep service business. It has become my ethical tech incubator, the umbrella under which I build any technology that involves other people's data or wellbeing. The Leadership AI Coach, Ray, and several builds outside this research all sit under DreamStorm. When I build under that name, the charter travels with the build.

Trurivu was the original name for the feedback app I thought I was going to build. I bought the domain. The Project Rise survey ran under it. When the focus shifted to conversational AI and vulnerable spaces, Trurivu went out of scope, but the Trurivu Kaupapa Charter remained, because the values it codified were not project-specific. They were mine.

The DreamStorm Charter has four core commitments (full charter and NO clauses in Appendix B1):

The charter is a living document. Cultural supervision from Lee, Nadine, and Rob shaped it in concrete ways. Lee's feedback on the "no memory" language (that people wouldn't understand what it meant) led to a clarification in all pilot materials: no conversation context carries between sessions. That shift came from a direct conversation and went straight into the charter language. Lee also surfaced the tapu/noa te reo Māori / Framework Terminology Core Value Tapu & Noa Tapu is the state of being sacred, restricted, under spiritual protection. Noa is the state of being ordinary, accessible. Research Context [2, 4, 8] ↗ "The central paradox of this research sits between them: AI risks making tapu things noa. But for people cut off from their culture, noa may be the only door available. That paradox is not resolved. It is held." tension for the first time, which didn't change the charter immediately but led to the Culture Meets AI wānanga te reo Māori Core Value Wānanga A gathering for deep learning and knowledge sharing. Research Context "In this research, the Culture Meets AI wānanga was a 90-minute session where participants explored together whether AI belongs in cultural spaces." : the charter asking a question it couldn't yet answer.

How to build yours

You don't need four builds and an 18-month masters to write a build code. You need honesty about what you actually value, and the willingness to name what you won't do before someone asks you to do it.

Start here:

draft_code.md

Draft Your Build Code

> Who is this actually for?

Not the business case. Not the stakeholder. The human being who will sit with whatever you build at 2am. Name them specifically.

> What do you believe about that person?

Write it down. "I believe they deserve X." If you can't finish that sentence, you're not ready to build for them yet.

> What will you never do?

This is harder than it sounds. Your NO clause needs to survive commercial pressure, tight deadlines, and a client who says "just this once."

> Who are you accountable to, culturally?

Name that accountability explicitly. Who do you check with? What does reciprocity look like?

> What does safety look like beyond a privacy policy?

A privacy policy is legal protection for you. Safety is relational protection for them. Name both.

> Who controls the data?

"We" is not an answer. Name the person. Name the system. Name what happens when you're no longer the builder.

> Where must a human stay?

Name those moments before you build, not after someone gets hurt.

"The things that align with your personal values most likely won't change. The things that need to adapt will tell you when."

The build code is a living document. Yours will change as your builds get harder and the stakes get higher. That's not a failure of the framework. That's the framework doing its job.

Cross-references

References