Skip to content

How it works · the same loop at every size

Write a decision into the brain once, and every Claude works from it.

Every business runs this loop, whether you run it yourselves or we run it with you. It takes 7 steps, and nothing that shapes how your team’s Claude answers changes until someone with authority says yes.

Every business runs this loop, whether you run it yourselves or we run it with you.

  1. 1Context
  2. 2Priorities
  3. 3Observe
  4. 4Decide
  5. 5Approve
  6. 6Write back
  7. 7Measure
See what the install covers

Step 1

First we spend 90 minutes with each person to build the brain

It builds one shared brain from what each person knows, in your own files and under your accounts. A check-in 2 weeks later makes sure it’s updating as it should. The rest of the loop reads from that brain and writes approved changes back into it.

It builds one shared brain from what each person knows, in your own files and under your accounts. A check-in 2 weeks later makes sure it’s updating as it should. The rest of the loop reads from that brain and writes approved changes back into it.

Your size decides who runs the loop

The install is the first thing we build at every size, and after that the size of your team decides who runs it. From 50 people, we usually start with the Decision Discipline Workshop.

Under 15 people

Self-directed between quarterly reviews

Your team runs the loop, and each team’s leader clears its shared suggestions each week.

Every person gets a monthly note, and a 1-hour review with us each quarter. The report card comes out of that review.

Context Install, then the Context Layer

15 to 49 people

We run the loop with the owner

Month 1 opens with a half-day priority session with the owner and up to 5 seniors.

Month 1 opens with a half-day priority session with the owner and up to 5 seniors.

Each month we bring a Deployment Brief to a 60 to 90 minute decision session with the owner, and whatever you approve is written back into the brain. Each quarter we review every initiative with you.

Adds Decision Deployment, per business, 12 months

Adds Decision Deployment, per business, 12 months

50 people plus

We run the loop with the executive team

The same loop runs department by department, with each department’s own owner and metrics.

Signals arrive weekly, and the executive team gets a monthly pack and a quarterly board review.

Adds the Decision Layer, per business, 12 months minimum

Adds the Decision Layer, per business, 12 months minimum

Every business starts with the Context Install. Each size up adds one thing to what the size below it has.

Add-on, any size

Decision Discipline Workshop

Two days in one room with the 5 to 15 people who can fund a decision and kill one. Every AI idea leaves funded, parked or killed, and each funded workflow gets an owner and a date.

It sits beside any of the three sizes. From 50 people it usually comes first, so the executive team agrees where AI goes before anything is built.

Priced per firm, for one room, on its own or beside the install.

Each step leaves a record you can open

StepWhat happensWhat it leaves behind
1ContextThe install captures how the business actually works.The shared brain in your files, and a Claude environment for each person.
2PrioritiesLeadership sets where AI should create value, what gets funded, who owns each initiative and when to stop.An owner for each initiative, and a baseline written down and dated.
3ObserveSignals come from the brain and the systems your work lives in.A monthly note for every person, and a monthly Deployment Brief when we run the loop with the owner.
4DecideYou make the calls on the evidence, on a fixed rhythm.Every call with an owner and a date.
5ApproveNothing changes without a person saying yes.The name of the person who said yes, and when.
6Write backApproved decisions land in the brain as updated context, workflows, instructions and skills.A decision record, and the updated file everyone reads.
7MeasureAgainst a baseline written down at the start.The quarterly report card.
  1. Context

    The install captures how the business actually works.

    What it leaves behind

    The shared brain in your files, and a Claude environment for each person.

  2. Priorities

    Leadership sets where AI should create value, what gets funded, who owns each initiative and when to stop.

    What it leaves behind

    An owner for each initiative, and a baseline written down and dated.

  3. Observe

    Signals come from the brain and the systems your work lives in.

    What it leaves behind

    A monthly note for every person, and a monthly Deployment Brief when we run the loop with the owner.

  4. Decide

    You make the calls on the evidence, on a fixed rhythm.

    What it leaves behind

    Every call with an owner and a date.

  5. Approve

    Nothing changes without a person saying yes.

    What it leaves behind

    The name of the person who said yes, and when.

  6. Write back

    Approved decisions land in the brain as updated context, workflows, instructions and skills.

    What it leaves behind

    A decision record, and the updated file everyone reads.

  7. Measure

    Against a baseline written down at the start.

    What it leaves behind

    The quarterly report card.

Then it repeats.

No AI spend without a funded workflow,
a named owner and kill criteria.

No AI spend without
a funded workflow,
a named owner
and kill criteria.

Stalled work shows up without anyone chasing or reporting it

Everyone reads the same brain, and leaders and managers also connect to the systems your work runs on. When we run the loop with you, we read the same brain and the same systems, read-only: mail, calendar, CRM, ledger and project tool. Never a staff member’s own inbox.

What it watches for

  1. Work that moves forward, and work that stops.
  2. Two people pulling against each other.
  3. Workflow automation opportunities, from what your CRM and other connected systems show.
  4. Good and bad news, and financial changes, as they were communicated.
  5. Anything left unactioned past the line you set.

Every person gets a monthly note on what moved, what stalled, what is waiting on that person and what has gone unactioned past 30 days. At 15 to 49 people, the owner also gets a monthly Deployment Brief, which adds what changed financially and what competitors actually did.

Monthly noteSeptember 2026

What changed in September

Prepared for the practice manager of a 12-person advice practice. An example.

Prepared for the practice manager of a 12-person advice practice. An example.

2
Waiting on you2 changes to approve before the October decision session.
What moved
The SOA preparation checklist is in use by every adviser.
What stalled
The fee template update has waited on finance since 2 September.
The fee template update has waited on finance since 2 September.
Unactioned past 30 days
An accountant’s referral from 18 August has had no reply.
An accountant’s referral from 18 August has had no reply.
Practice managerFor the October decision session

An example monthly note from the same 12-person advice practice as the homepage, which runs its own monthly decision session. Everything in it is illustrative.

Monthly noteSeptember 2026

What changed in September

Prepared for the practice manager of a 12-person advice practice. An example.

Prepared for the practice manager of a 12-person advice practice. An example.

2
Waiting on you2 changes to approve before the October decision session.
What moved
The SOA preparation checklist is in use by every adviser.
What stalled
The fee template update has waited on finance since 2 September.
The fee template update has waited on finance since 2 September.
Unactioned past 30 days
An accountant’s referral from 18 August has had no reply.
An accountant’s referral from 18 August has had no reply.
Practice managerFor the October decision session

An example monthly note from the same 12-person advice practice as the homepage, which runs its own monthly decision session. Everything in it is illustrative.

We never read your team’s chats

We never read what anyone
types into their own Claude.

Nothing on the list above needs it. Staff get this line in writing at every install.

Each quarter, the report card shows how every person is doing, from the date of their last update to what’s waiting for their yes. Everyone sees what it says about them.

Once a decision is written where the team reads it, it stops coming back

Nothing that shapes how your team’s Claude answers changes until someone with authority approves it. The record names who approved it and when.

Each person approves changes to their own files. Changes to the shared brain go to the leader who manages that team, and anything that crosses teams goes up the chain of command.

A fact picked up from email, documents, meeting transcripts or your systems waits in a suggestion inbox before it can reach a file your team’s Claude reads at the start of every chat. It carries the source, the date, the exact quote and the file it would change. Anything arriving that way can suggest a fact but never set a rule.

Waiting on you

Proposed change

Client reviews move to 2 meetings

From
October decision session
Owner
Practice manager
Changes
client-review.md
review-letter.md
Reaches
12 people
ApproveSend back

Proposed change

Client reviews move to 2 meetings

From
October decision session
Owner
Practice manager
Changes
client-review.md
review-letter.md
Reaches
12 people
ApproveSend back

Approved 9 October by the practice manager

Decision record

Client reviews move to 2 meetings

Approved 9 October by the practice manager

From
October decision session
Owner
Practice manager
Written into
processes/client-review.md,
templates/review-letter.md
Reaches
12 people

Decision record

Client reviews move to 2 meetings

Approved 9 October by the practice manager

From
October decision session
Owner
Practice manager
Written into
processes/client-review.md,
templates/review-letter.md
Reaches
12 people

The same example practice as the homepage. File names, roles and dates are illustrative.

The target is 5 hours saved a week per person, checked each quarter

The target is 5 hours saved per person, per week. The quarterly report card shows whether you reached it.

The install writes down the recurring tasks the hours are measured against, how long each takes today and how often it comes up, and you sign that list. At each quarterly review you estimate the same tasks the same way. Only confirmed tasks count, and every input sits beside the result.

In month 1 of running the loop with the owner, we write down the 3 to 5 numbers the year is measured against. Each quarter, every initiative is marked fix, fund, scale or stop, in a pack you can hand to your board.

In month 1 of running the loop with the owner, we write down the 3 to 5 numbers the year is measured against. Each quarter, every initiative is marked fix, fund, scale or stop, in a pack you can hand to your board or your licensee.

Report card
Quarterly report cardSample data: Northwind Advice is fictional

Northwind Advice, Q3 2026

Quarter to date, 1 Jul to 22 Sep. The business brain and 4 people, measured against the baseline signed on 30 June: 6 named tasks, estimated the same way both times.

Client
Northwind Advice
Period
Q3 2026, to 22 Sep
Exported
22 Sep 2026
File
pulse-NWA-2026-Q3.json
169.2 hTime saved, confirmed
Target 240 h
6Corrections this quarter
9 of 13Known gaps closed
3 of 4People writing back

Where each person stands

PersonStatusLast dated entryHours a week
SamPrincipalHealthy1 day ago5.5 h
LeeAdviserHealthy4 days ago5.2 h
AnaParaplannerWatchToday3.4 h
RajOffshore adminQuiet26 days ago0 h

What the numbers suggest

1

Check Raj’s capture before anything else

Last dated entry 27 Aug, 26 days ago, past the 21-day quiet line.

Suggested move: book 15 minutes with Raj. The laptop is personal, so Raj makes the fix with you on the call.

Page 1 of 4Read from pulse-NWA-2026-Q3.json

A quarterly report card, shown with sample data. Northwind Advice is fictional. Each quarter adds a card to the series.

Report card
Quarterly report cardSample data: Northwind Advice is fictional

Northwind Advice, Q3 2026

Quarter to date, 1 Jul to 22 Sep. The business brain and 4 people, measured against the baseline signed on 30 June: 6 named tasks, estimated the same way both times.

Target 240 h
169.2 hTime saved, confirmed
6Corrections this quarter
9 of 13Known gaps closed
3 of 4People writing back

Where each person stands

PersonStatusLast dated entryHours a week
SamPrincipalHealthy1 day ago5.5 h
LeeAdviserHealthy4 days ago5.2 h
AnaParaplannerWatchToday3.4 h
RajOffshore adminQuiet26 days ago0 h

What the numbers suggest

1

Check Raj’s capture before anything else

Last dated entry 27 Aug, 26 days ago, past the 21-day quiet line.

Suggested move: book 15 minutes with Raj. The laptop is personal, so Raj makes the fix with you on the call.

A quarterly report card, shown with sample data. Northwind Advice is fictional. Each quarter adds a card to the series.

Questions about
running the loop

It stays where it is, in your own storage and under your own accounts. The context files, skills, configuration and documentation written for your business are yours once the install is paid in full, to use inside the business, and stopping the ongoing work doesn’t change that. We keep the rights to the method and templates they’re built from.

Joel HauerFounder, MethodWorks

Book a call with Joel

Joel and the MethodWorks team put AI into the daily work of Australian businesses in financial advice, retail, logistics, education and software. Bring your headcount and the AI tools your team uses today, and you’ll leave the call knowing who your Context Install would cover.