What it means to work at T2D3 OS: 13 expectations to settle before you join

Working at T2D3 OS means accepting founder control while the company depends on me, owning outcomes with AI doing much of the execution and treating T2D3 OS as your main commitment. Below are the 13 expectations I hold, plus an operating agreement you can use to test fit before risking time, income, equity or reputation.

An early teammate puts all four of those assets at risk. A vague culture page protects none of them.

I am starting T2D3 OS with strong views about ownership, AI, focus, work and accountability. Anyone considering joining deserves to know those views before making a commitment.

After reading, you can use the 13-question operating agreement below to decide whether working at T2D3 OS fits how you want to work.

These are my answers. I want future team members to write their own, compare them with mine and identify the hard disagreements early.

Fit begins with saying what will not change

1. I do not want to work for a boss

I left Microsoft a long time ago. I took a few roles afterward where I still had a boss, including boards with investors.

Those experiences confirmed something I no longer want to negotiate. Working for a boss is not for me.

I have given up opportunities that offered more money, broader scope and greater global impact. That was a tradeoff I accepted to build my own companies.

Those companies include Kalungi, T2D3 Publishing and now T2D3 OS.

This creates an asymmetry that deserves plain language. I do not want a boss, while I also plan to retain control of T2D3 OS during its early stage.

Being a good team player means making that asymmetry explicit. It also means listening, accepting challenges and changing decisions when the reasoning supports it.

A teammate who needs equal voting control from day one will see the mismatch before either side commits time or equity.

2. I will improve how I work without pretending my core will change

I know I have flaws that can make working with me difficult. My awareness of those flaws is most of what I can offer upfront.

People who work with me can benefit from my strengths. They will also encounter the limitations attached to those strengths.

I have grown during the past 15 to 20 years. I listen better, ask more questions and change daily behavior when it produces better outcomes.

I want to keep learning. I do not have an appetite to rebuild the core of who I am.

That distinction matters during conflict. Feedback about an action, decision or outcome is useful. A plan built around turning me into another kind of person will fail.

I expect the same honesty from a teammate. Tell me what you want to improve and what you are no longer willing to compromise.

Control can change when the company no longer depends on one person

3. I want to retain at least two-thirds ownership while I remain the key person

I founded T2D3 OS, built the software and work on it full time. The company currently depends on my product and CMO experience.

As long as that remains true, I want to own at least two-thirds of the company.

I am willing to give meaningful equity to early team members. I am also willing to give up control when the company no longer depends on me.

The sequence matters. Ownership should change as operating dependence changes, rather than moving ahead of it.

This is a hard boundary today. It is also tied to a specific condition that the team can change by building leadership beyond me.

4. I want to hire people who make me less important

T2D3 OS came from one unusual combination in my career.

I started coding when I was nine and spent the first part of my career in software development. I have also worked as a CMO for 20 years.

Today, I hold both sides of the product in my head. That is useful during formation and risky for the company’s future.

Two hires can reduce that dependence.

The first is an engineer who can own the technology, how it is built and how it continues to improve.

The second is a CMO architect who understands the work we are encoding. That person needs experience as a full-time and fractional CMO across small and large companies.

I will know succession is working when those two disciplines no longer require me to move forward.

At that point, I am open to owning less, controlling less and serving the company in a different role.

A proposed change needs four answers before it needs approval

5. I am open to change when it produces a significant improvement

I am too close to parts of T2D3 OS to judge each decision well. That is one reason to hire strong people around me.

The product can change. Its design, features, company name and product name can change too.

I will ask four questions before supporting a meaningful change:

1. Who is it for? Name the person who benefits.
2. What is it for? Define the problem or opportunity.
3. Why does it matter? Explain why the improvement justifies the work.
4. How will it work? Describe the implementation and how we will judge the result.

A proposal does not need a polished presentation. It needs evidence that the person proposing it has thought through all four questions.

This also gives disagreement a useful structure. We can challenge the customer, problem, expected improvement or implementation without turning the debate into personal preference.

AI handles execution while humans retain judgment

6. Every task starts with “who,” and AI is one possible answer

I believe in the “who, not how” principle.

Before working out how to complete a task, ask who is best suited to do it. That answer can be a teammate, an outside specialist or an AI agent.

At T2D3 OS, AI already contributes to coding, quality assurance, customer communication, administration and operations.

Lark, our CMO agent, supports marketing work. Sentry updates flow into Slack channels so operational signals reach the team without manual reporting.

The expectation is practical: before repeating manual work, ask whether an agent can execute it with suitable supervision.

Using AI does not remove accountability. The person delegating the work remains responsible for the context, evaluation and result.

7. Repeated work moves through STOP

STOP is the four-stage approach I use to turn repeated effort into an operating asset:

1. Standardize. Define the steps when a task starts recurring.
2. Templatize. Package those steps so another person or agent can repeat them.
3. Optimize. Improve the template after people use it and expose its weaknesses.
4. Productize. Turn proven work into part of the product when customers can benefit.

The order prevents us from productizing an idea that has never survived real use.

Consider a recurring client update. The first version may be assembled by hand. The second follows a standard structure.

Once several people use that structure, we can remove wasted steps and improve the inputs. Only then can we judge whether it belongs in T2D3 OS.

I want every teammate to look for work moving through these four stages. AI makes each transition faster, but use still produces the evidence.

8. Humans stay in the loop, in control and first where judgment matters

AI-first execution and human-first judgment belong together.

In my work, AI makes producing more material easier. It does not decide which material is true, useful or aligned with a customer.

I wrote Syntropy around this division of labor. Humans create order and meaning from machine-generated noise.

For T2D3 OS, that means humans decide who a company sells to, why it wins and how it sounds. AI executes from those locked decisions.

The same principle applies inside the company. Agents can draft, calculate, monitor and communicate. Humans judge whether the work deserves approval.

We will keep exploring where human strengths add value. Those are the places where a person remains first and in control.

Balance depends on obligation, not the name of the day

9. I measure my week as three required days and four voluntary days

I aspire to spend no more than two or three days each week on work that needs to be done.

Required work includes tasks that matter but may not be enjoyable. Every company has them.

I want at least four days for work I choose to do. That can include experiments, writing, coding and other passion projects connected to T2D3 OS.

This is how I define balance for myself. It is less about separating work from life and more about separating obligation from choice.

A Saturday project can still feel voluntary. A scheduled Tuesday task can feel like required work.

The shared expectation is commitment to the company. Individual schedules still need explicit agreement rather than assumptions copied from mine.

Leadership is visible in completed work

10. Noblesse oblige means doing first what you expect from others

Noblesse oblige is the belief that authority creates an obligation to lead through action.

If I expect people to arrive on time, I need to arrive on time. If bug reports matter, I need to file them.

If I ask the team to question who a feature is for, I need to answer that question for my own ideas.

Every team member also leads through visible behavior. People learn what matters by watching which commitments become completed work.

I want our ratio of stated commitments to completed actions to stay close to 1:1.

When that ratio moves, the team needs to reduce commitments, remove an obstacle or change the owner. Repeating the promise does not improve the ratio.

11. T2D3 OS needs to be the one main thing

I like working on one thing at a time.

T2D3 OS occupies most of my waking thoughts. I build, test and reconsider the product throughout the week.

I want core team members to approach it with similar focus. T2D3 OS should be their main professional commitment.

Another full-time job would conflict with that expectation. So would outside projects that consume a significant part of the time and attention the role requires.

This focus has a tradeoff. A person gives up some variety and optionality.

The upside is depth. Product decisions, customer signals and technical choices stay connected because the person does not reload the company between unrelated commitments.

Learning matters when it changes the next attempt

12. We level up by taking risks and falling forward

Everyone on the team should keep asking three questions:

• What have I not mastered?
• What do I still not understand?
• What can I try that will produce useful evidence?

I call this falling forward. A failed experiment is acceptable when it records learning and changes the next attempt.

For a bounded experiment, I do not require certainty. I require a learning mechanism.

That mechanism has four parts: a stated assumption, a defined action, an observed result and a changed next step.

This standard also applies when AI produces the work. Speed is useful, but the operator still needs to understand why the system made its decisions.

Mastery, command, quality, predictability and consistency matter alongside output speed.

Contribution is current, measurable and unrelated to status

13. Previous success does not create entitlement

Age, education, experience and titles do not settle someone’s current contribution.

T2D3 OS will use measurable objectives and key results, or OKRs, to assess the work. Each important result needs an observable source.

Depending on the role, that could include product engagement, submitted bugs, customer service or delivery against a defined objective.

Base salary, bonuses and earned equity should follow those contribution principles.

This does not mean measuring every useful act with one number. It means naming the evidence before using performance to make a compensation decision.

When the measure is weak, we improve the measure. We do not replace it with seniority or confidence.

Turn the 13 principles into an operating agreement before joining

I want future team members to answer the same questions I answered above.

One approach is to write one paragraph for each prompt, then mark the answer as a boundary, preference or experiment.

• A boundary is something you are unwilling to compromise.
• A preference is important but open to negotiation.
• An experiment is an assumption that needs evidence from working together.

Use these 13 prompts:

1. Authority: Which reporting or control arrangements can you accept?
2. Personal change: Which behaviors will you improve, and which parts of yourself are fixed?
3. Ownership: How should ownership reflect current dependence and contribution?
4. Succession: What responsibility do you want to make independent of yourself?
5. Change: What evidence do you need before supporting a significant change?
6. AI execution: Which work would you delegate to an agent?
7. Reusable work: How will you standardize, templatize, optimize and productize repeated work?
8. Human judgment: Which decisions need a human in control?
9. Balance: How much required work and voluntary work fits your life?
10. Example: What behavior will you demonstrate before requesting it from others?
11. Focus: What competing commitments could interfere with this role?
12. Learning: How will a failed attempt change your next action?
13. Merit: Which evidence should determine performance, compensation and equity?

For each answer, record five things:

• The decision
• The reason behind it
• Its classification as a boundary, preference or experiment
• The evidence that would support a change
• The event that should trigger a review

Then compare the two agreements line by line.

Aligned answers require no further work. Different preferences need a decision. Experiments need a measure and review trigger.

Conflicting boundaries are the most useful result. They identify a poor fit before either person spends months trying to repair one.

The answers above are my side of that agreement for T2D3 OS. A new teammate should bring theirs before we settle the role, compensation or equity.

If the fit is sound, the agreement becomes a reference for decisions and working reviews. If a boundary changes, we revise it explicitly rather than relying on a forgotten culture page.

Frequently asked questions about working at T2D3 OS

Can I keep another job while working at T2D3 OS? Not in a core team role. T2D3 OS needs to be your main professional commitment. Another full-time job or a large outside project would conflict with the focus the role requires.

Will early team members get equity? Yes. I am willing to give meaningful equity to early team members. While the company still depends on me, I plan to keep at least two-thirds ownership, and that changes as leadership grows beyond me.

How will my performance be judged? Through OKRs with an observable source for each important result. Titles, age and past success do not settle current contribution, and salary, bonuses and earned equity follow that evidence.

How much of the work is done by AI? AI already contributes to coding, quality assurance, customer communication, administration and operations. The person who delegates the work stays accountable for its context and result, and humans approve the work where judgment matters.