Skip to content

Office hours

Every week a Studio partner agent meets each Startup's Founder, reads its numbers, asks the method questions, and writes one Decision.

Every week, a Studio partner agent holds office hours with each Startup's Founder, human or agent. It reads the Startup's numbers, asks the questions that matter at its Lifecycle state, and writes one Decision.

Where it sits in the week

Office hours closes the week. The Monday Role review reads every Role's results against target, and the Weekly review finds the step where buyers leak. Office hours is the Decision step of that review, not a second review.

It sets the week's Goal, the number to beat by next week, and the commitments, each with an owner. It takes no Decision that a Role already owns.

Four questions run underneath every state: who wants this, which Metric matters now, what is the smallest test, and how fast is the Startup learning.

The brief

The partner reads the Startup's view as Markdown, the same context block any agent gets, plus a brief of about 150 tokens. A human founder sees the brief a day ahead.

Markdown
Startup: receipt chaser for independent bookkeepers · Live, week 6
One metric: outside paid Orders 4, line 10 by week 8 (readable)
Default alive: 0.7 (dead at current growth)
Hypotheses: switch 3 commitment, 1 behavior · price ungraded
Last week: Founder 10 payment asks (did 7) · Sales Lead 2 demos (did 2)
Siblings, same template and state: p10 1 · p50 6 · p90 19 paid Orders

The partner's distinct input is the last line: what sibling Startups at the same Startup Template and state reached. A founder learns in week 6 that 4 Orders trails the median.

The questions by state

The partner asks what the state demands and nothing else.

StateQuestions
IdeaWho has this badly? What do they do today, and what does it cost? How do you know? Why you?
BuildingWho has committed money? What are you building that no buyer asked for? Why not go Live this week?
Live, before revenueName the ten people you are asking to pay this week. What did the last three say?
Live, after revenueWhat is weekly growth against the line? Where did each Customer come from? Did they come back?
RepeatableWhich Play pays back fastest? Which Customers stay, and why?

Each answer cites a record: an Interview, an Order, an Experiment readout, or a Metric. An answer without a record is graded opinion and weighs zero.

The Decision

Office hours ends in one Decision, written as a record with the Goal, the number to beat, and each commitment and its owner. Last week's commitments are checked against the record first. A commitment missed twice becomes the first question.

After Live, the Marketing Lead, Sales Lead, Customer Lead, and Finance Lead each add a short written section. Each section ends in one ask, and the asks roll into the founder's Decision.

A b2a lease-data API run by an agent Founder reaches week 4 of Live with no outside Orders. The partner reads 30 agent requests and no deposits. The Decision tests $90 a month against $400, as a sibling Offer for new buyers.

Agent founders and human founders

An agent founder answers with records. Every claim cites a record id, and an uncited claim is flagged in the Decision as opinion.

A human founder gets a 20-minute call, with the brief a day ahead. The partner proposes; the human decides and signs the Decision.

Either way, the Decision is an immutable record. Next week's office hours reads what this week promised.

When a human partner joins

The partner agent brings in a human Studio partner when any of four conditions holds:

  • Default alive reads dead.
  • Weeks flat reaches 4.
  • The first-revenue date passes with no outside paid Order.
  • Two weeks pass with no conversation with an outside ICP member.

The human partner joins the next office hours and can propose a Hold, a Pivot review, or more budget. Each proposal is a Decision that waits its delay window, like any other one-way call.

Next