Skip to content

How to operate at Duvo

Duvo gives people room to act and expects them to own the result. If you manage someone, also read the Manager Playbook.

Why we work this way

Duvo works on consequential problems inside large enterprises. A better decision can unlock millions; unreliable execution can stop work or cost revenue. We need startup speed and enterprise-grade dependability at the same time.

We work in public, default to async, and let the person closest to the problem make the call. We agree on the outcome, then use agents to get there faster.

Why Duvo is an awesome place to work!

You will learn how large enterprises really work by getting close to problems in finance, supply chain, and operations. You will see how experienced operators think, how customer knowledge becomes a product other companies can use, and why dependable execution matters at scale.

You will learn to make agents part of your craft: finding context, creating the first useful version, testing ideas, and improving your own work. We share what works so the whole company gets better faster, while each person remains responsible for the judgment and result.

You are trusted with real responsibility from the moment you join. The work will stretch your craft and judgment, and you will get direct feedback while it can still help you improve. As Duvo grows, new teams and roles will emerge. We want people already here to grow into them, with clear guidance on what the next level requires.

What Duvo expects from you

In return, Duvo expects you to:

  • Start before you have perfect certainty.
  • Share work and reasoning where others can see them.
  • Use agents to improve the speed and quality of the work.
  • Ask for and act on direct feedback.
  • Keep your commitments or renegotiate them early.
  • Stay with the work until the intended outcome is real.
  • Leave useful learning in the company's shared context.

Move quickly on decisions that are easy to reverse, contained, and within our agreed direction. Escalate before decisions that are hard or costly to reverse, set a strategic precedent, make a material external commitment, or create serious people, legal, security, financial, or reputational risk.

With customers, improve the process before automating it. Tell them when something is not worth fixing.

The default operating loop

Start with the outcomeDescribe what should change for a customer or Duvo and what evidence would prove it.

Find the existing contextCheck source systems, previous decisions, customer evidence, and the current owner before asking someone to reconstruct the history.

Create the first useful versionDo the work yourself or give a bounded first pass to an agent.

Share it earlyPut the work, assumptions, and open questions in the relevant public channel while feedback can still change the direction.

Use the feedbackResolve disagreement with evidence, make trade-offs explicit, and improve the work.

Verify the resultCheck the actual customer, product, commercial, or operational state. Activity alone does not prove completion.

Leave reusable contextRecord the decision, correct outdated information, and make repeated work easier next time.

Run this loop in small steps. Early visibility makes wrong assumptions cheaper to correct.

The bar

This is who we hire and keep. Good and bad are shown together so the line stays visible.

We judge the work on customer outcomes, focus and judgment, dependable delivery, role craft, and whether you raise the people around you.

1. Builds from zero

GoodHas personally created products, systems, processes, or commercial motions without relying on a mature playbook, brand, or support structure.

BadResults only exist inside a scaled organisation where the processes, team, and resources were already in place.

2. Shows high agency and ownership

GoodActs without waiting for perfect instructions, stays close to the work, and owns outcomes through problems and iteration.

BadCoordinates, advises, or delegates by default, and blames another team, the customer, or missing processes when results fall short.

3. Obsesses over customers and measurable outcomes

GoodConnects their work to a real customer or business result and can explain the evidence clearly.

BadTalks mainly about activities, outputs, prestige, or internal processes without a clear outcome.

4. Is pro-AI and AI-native

GoodUses agents as normal leverage, can show how they change the speed, quality, or scope of the work, and still owns the judgment and result.

BadTreats AI as optional, hands automatable work to colleagues, or cannot explain and verify the output.

5. Chooses Duvo deliberately

GoodUnderstands the stage, wants the responsibility, and chooses the work with open eyes.

BadWants the title or package without the responsibility and ambiguity that come with the stage.

If this page is wrong, say so. We would rather fix it than let folklore win.