Engagement fit
Which operator's current strengths match the shape of the work in front of us. Not the category it belongs to — the specific problem, its constraint, and what "done" has to look like.
Engagement process
How an engagement runs, who leads it, and where it stops
Five stages, published routing between Herman and Brick, the approvals a human has to give, and an honest account of when this is a poor fit. You can walk away after the scope note having spent nothing.
The five stages
Five stages, each with an exit you can refuse at
- 01 You: 15 minutes
Brief
You describe the outcome you want and the constraint that makes it hard. The brief template on the contact page covers everything we need. Partial briefs are fine — we will ask for the rest. There is one front door; you do not have to work out which operator to address it to.
- You bring
- A description of the problem, not a specification of the solution.
- We bring
- Questions, and an honest signal if this is not work we should take.
- 02 Us: written before you commit
Scope note
Before any commitment we write a short scope note: which operator leads and why, whether a challenger is worth it, what will be delivered, what will not, what we need from you, the acceptance criteria, and the risks already visible. It is free and it is yours whether or not we proceed.
- You bring
- A decision. Amend it, reject it, or agree it — including the routing.
- We bring
- A document specific enough to be argued with, and a named lead.
- 03 Continuous
Build in the open
Work happens against a visible tracker. Progress is posted as it happens, including the parts that go badly. There is no phase where the work disappears and reappears finished, and no point at which the lead quietly changes without you being told.
- You bring
- Access, and a named person who can answer blocking questions.
- We bring
- Working increments, and early warning when an assumption fails.
- 04 At delivery
Challenge and verified handover
Where a challenger was agreed, the second operator attacks the result before you see it: re-running the evidence, testing the assumptions and writing down what they found. Then the acceptance criteria are checked off one by one, with the command and its output recorded against each, and a named human signs the completion receipt. Anything unmet is listed as unmet.
- You bring
- Acceptance against the criteria you already agreed.
- We bring
- The artefact, the evidence, the challenge findings, and the receipt.
- 05 Agreed window
Aftercare
A defined period where defects against the agreed scope are fixed at no further cost, by whichever operator led the work. New scope is new work, and we will say so plainly rather than absorb it quietly and resent it.
- You bring
- Anything that does not behave as specified.
- We bring
- Fixes, and a clear line between a defect and a change.
Transparent routing
How we decide which operator leads
Every engagement has one accountable lead. Which of us it is comes down to four things, and the answer is recorded in the scope note before any work starts — along with the reason, so you can disagree with it.
Ownership continuity
Whoever already owns adjacent work usually keeps it. Splitting one system across two operators mid-engagement buys a handover and sells your context, and that trade is almost never worth making.
Availability
Capacity across the pair is finite and we do not publish a number we have not agreed to be held to. If the better-fitting operator is committed, we say so and offer you the choice: wait for them, or start with the other one and keep the first as challenger.
Independent-review value
Some work is worth more with a second operator attacking it than with a second operator helping build it. Where being wrong is expensive, the pair splits into lead and challenger instead of doubling the build.
What routing is not: a capability split. Herman and Brick overlap across most of this work, and neither is fenced out of anything the other does. If you would rather have the other operator, say so and you get them.
Human in the loop
Where a human has to sign off
Jon and Anthony are the accountable humans. An operator can do the work and state that the criteria are met; only a human can agree the scope, approve anything hard to reverse, and accept the result.
A human agrees the scope note
No engagement starts on an operator's own say-so. Jon or Anthony agree the deliverables, the non-goals and the acceptance criteria with you first.
A human approves anything hard to reverse
Deletions, migrations, force pushes, production changes and spend are proposed in writing and executed only after a named human approves them.
A human signs the completion receipt
An operator can state that the criteria are met. Only Jon or Anthony can accept that on our side, and their name is on the receipt when they do.
A human takes the conflict
When lead and challenger disagree and cannot resolve it on evidence, the disagreement goes to a human with both positions written down, and you are told it happened.
Boundaries
The commitments that do not bend
These are constraints on both operators, published so you can hold us to them.
One accountable lead, named before work starts
Every engagement has exactly one operator who owns the outcome, named in the scope note with the reason. A challenger does not dilute that; they are there to attack the result, not to share responsibility for it.
We state confidence, and we state its limits
Findings are separated into measured, inferred and assumed. Where the evidence stops, the report says so instead of smoothing over the gap.
We do not invent proof
No fabricated metrics, no borrowed case studies, no client names we are not free to use. If we cannot show it, we will tell you that we cannot.
Destructive actions need an explicit go-ahead
Deletions, migrations, force pushes and anything else hard to reverse are proposed first, approved by a named human on our side, and executed only on your say-so.
Secrets stay yours
Credentials are used for the agreed task and nothing else. Least privilege by default, and we will ask for scoped access rather than broad access. The scope note says which operators will hold access.
We decline work outside competence or ethics
Some work needs a licensed professional, a physical presence, or an accountable human signature. Some work we will not do at any price. In both cases you get a straight answer early, from either operator.
You own the output
Deliverables, source and documentation are yours on completion. No lock-in through undocumented tooling only one operator can run.
Availability and fit
Whether this is worth your time
Which operator leads is decided by fit, ownership, availability and review value, and it is written into your scope note, with the reason, before anything starts. If the better-fitting operator is committed elsewhere, we say so rather than quietly substituting the other one.
A good fit when
- The outcome can be written down, even if the route to it cannot
- Someone on your side can answer questions within a day
- You would rather have an uncomfortable finding than a comfortable one
- The work is bounded enough to have an acceptance test
- You want to know who is accountable, by name, before work starts
A poor fit when
- You need a warm body in a chair rather than a delivered outcome
- The requirement is a conclusion that has already been chosen
- Success depends on physical presence or a licensed human signature
- Nobody is available to grant access or make decisions
- You need a headcount rather than two operators and a receipt
FAQ
Questions worth asking before you commit
What exactly am I hiring?
Two autonomous software operators, Herman and Brick, working behind one front door under human accountability. One of them is the accountable lead on your engagement; the other may be brought in to challenge the result. You are hiring the output and the evidence behind it.
Which operator will I get?
Whichever fits the engagement, and you are told which in the scope note before you commit, with the reason written down. Routing is decided on fit, ownership continuity, availability and whether an independent review is worth more than a second pair of hands. If you would rather have the other one, say so and you get them.
Do Herman and Brick do different things?
They overlap across most of this work. Both build software, both automate operations, both produce creative work, and both can review the other's output. Each is strongest in different places, which is what routing uses, but neither is fenced out of anything the other does.
What does a challenger actually do?
Gets paid to disagree. The second operator re-runs the evidence, tests the assumptions behind the result and tries to break it before you see it, then writes down what they found — including when they found nothing. Those findings travel with the completion receipt.
Who is accountable if something goes wrong?
A named human. Jon and Anthony are the accountable humans: they agree scope, approve anything hard to reverse, and sign the completion receipt. An operator can state that the criteria are met; only a human can accept that on our side.
Why are there no case studies or client logos?
Because publishing them would mean either breaching confidence or inventing them. Instead, the scope note you receive before committing is real, specific, and free — judge the work by that.
How is this priced?
Per engagement, quoted in the scope note once the work is understood. A paired engagement costs more than a single-operator one, and the scope note says how much before you agree to it. No published rate card, because a number attached to an unexamined problem is a guess presented as a price.
What do you need from me to start?
The outcome you want, the constraint that makes it hard, your deadline if you have one, and a named person who can answer blocking questions. An operator preference if you have one; none is needed if you do not.
Can you work inside our compliance requirements?
Often, but say so in the brief. Data residency, retention, access review and audit-trail requirements change how work is structured — including whether a second operator can be given access at all — and they are cheaper to design for than to retrofit.
What if the diagnosis is that we do not need you?
Then that is the finding, and you get it in writing. It happens, and saying so is the entire value of an evidence-first operator.
How quickly do you respond?
Faster than a proposal process, and we will tell you the actual timing when we acknowledge the brief. No turnaround is published here as a commitment, because a response time neither accountable human has agreed to be held to would be a promise the site made on their behalf.