Skip to content

INITE Protocol: six stages from brief to a system that runs

Most companies are quoted six to twelve months for this kind of work. Break through Cast takes twelve weeks, and this page is the reason we can promise that: every stage has a fixed end, something you receive, and a decision point where you can stop. Form is the sixth stage and it is optional — it runs from month three onward, and nothing stops working if you decline it.

The six stages12 · weeks to everything running
  1. 01Week 1–2

    Break

    Diagnose & Reset

    What you get

    A process map with the bottlenecks marked

    Before we go on
  2. 02Week 2–3

    Hold

    Stabilize & Fix

    What you get

    Written procedures for the workflows that matter

    Before we go on
  3. 03Week 3–4

    Track

    Measure & Analyze

    What you get

    A live dashboard of the numbers that matter

    Before we go on
  4. 04Month 2

    Cut

    Simplify & Eliminate

    What you get

    Streamlined process flows

    Before we go on
  5. 05Month 2–3

    Cast

    Build & Deploy

    What you get

    1–3 automated processes doing real work in production

    Before we go on
  6. 06Month 3–6

    Form

    Optimize & Scale

    What you get

    A monthly written report on how it is doing

    If you decline
6 · stages, each with a clear end1–3 · processes live in the first delivery

12

weeks to everything running

6

stages, each with a clear end

99.9%

uptime, written into the contract

1–3

processes live in the first delivery

Why this kind of project usually fails

The reasons repeat, and none of them are technical. The scope keeps growing. Nobody agreed what success looks like. The data turns out not to be there. The people who will use it were not asked. And the testing happens at the end, when it is too late to change anything.

The usual way

  • Six to twelve months before anything is used
  • Budget ends up two to three times the estimate
  • What is being built is still unclear halfway through
  • Everything is tested at the end
  • The people who will use it are not in the room
  • Most of it never goes live

How we do it

  • Twelve weeks to everything running
  • Fixed price agreed up front
  • Scope locked in week two and written down
  • Something checkable every two weeks
  • A conversation with your team every week
  • You can stop at the end of any stage

The six stages

The six stages12 · weeks to everything running
WEEK04812162024BreakHoldTrackCutCastForm12 · to everything runningForm · optional

Week 1–2

Break — Diagnose & Reset

We follow the work as it actually runs today, put a number on what the manual version costs, and decide together what is worth automating first.

What happens

  • We follow the work as it really moves through the company, not as the org chart says
  • We talk to the people who do it every day
  • We check the data exists and is good enough to build on
  • We put a number on what the manual version costs you now
  • We agree what counts as success, in numbers

What you get

  • A process map with the bottlenecks marked
  • What the current way costs, in hours and money
  • A priority order: what to automate first, and why
  • A written scope, priced
  • The numbers we will be judged on

Before we go on

Everyone signs off on the scope. Nothing gets built until that is in writing.

Week 2–3

Hold — Stabilize & Fix

We fix the process before automating it. Automating a broken process gets you a faster broken process.

What happens

  • We write down how the critical work is supposed to run
  • We clean up the data the automation will depend on
  • We make the manual fixes that pay off immediately
  • We remove the steps that only exist because something else broke

What you get

  • Written procedures for the workflows that matter
  • Data cleaned up where it was going to bite
  • Manual fixes already saving time, before any automation

Before we go on

The process runs the same way twice in a row. Until it does, there is nothing stable to automate.

Week 3–4

Track — Measure & Analyze

We put measurement inside the processes and let two weeks of real numbers say which ones are worth automating.

What happens

  • We instrument the processes themselves
  • We collect real numbers instead of estimates
  • We look for the patterns: where the time goes, where things stall
  • We score each process for how well automation would actually work on it

What you get

  • A live dashboard of the numbers that matter
  • What the data says about where the time goes
  • A feasibility score per process, with the reasoning

Before we go on

The baseline is agreed. Every improvement from here is measured against a number both sides accepted.

Month 2

Cut — Simplify & Eliminate

We take out the steps that exist for no reason, before anyone writes code around them.

What happens

  • We remove steps that no longer have a purpose
  • We merge processes that were doing the same job twice
  • We reshape what is left so it can be automated cleanly
  • We write the specification the build will follow

What you get

  • Streamlined process flows
  • The redundant steps gone, listed so you can see exactly what went
  • Specifications ready to build against

Before we go on

You approve what was removed. Anything you want kept, stays.

Month 2–3

Cast — Build & Deploy

The automation goes into production, connected to the tools your team already uses. Not a demo environment.

What happens

  • We build the part that does the actual work
  • We connect it to your CRM and the rest of your systems
  • We connect the channels your customers actually write to
  • We move the real flow across a bit at a time
  • We train your team and check they can run it
  • We set up the alerts that tell you when something is wrong

What you get

  • 1–3 automated processes doing real work in production
  • Integration with the tools you already have
  • A screen showing what is happening right now
  • Written instructions, and people on your side who can run it
  • A plan for what to do when something breaks

Before we go on

Everything is live and watched. This is the point at which it is yours.

Month 3–6

Form — Optimize & Scale

Optional, and priced separately. Nothing stops working if you decline it.

What happens

  • We watch how it performs, day and night
  • We tune it on what real usage shows
  • We try improvements against the current version and keep what wins
  • We feed in what your people report
  • We plan the next wave of processes
  • We sit down with you every quarter

What you get

  • A monthly written report on how it is doing
  • Improvements based on real usage, not guesses
  • A plan for the processes that come next
  • Someone to call

If you decline

If you decline: the system keeps running and the code is already yours. You can hire anyone to look after it.

What we guarantee in writing

  • 99.9% of the time it is working — standard
  • 99.95% of the time it is working — for larger installations
  • A fault fixed within 24 hours
  • A written report every month

What you can hold us to

12
weeks, brief to running system
every stage has a fixed end and a decision point
24h
to a fault fixed
written into the contract, not a target
0
lock-in
the code is yours from the first stage

From one process to the whole company

Getting one process working is the easy part. Most companies stall on the second one — not because the idea failed, but because what carried a hundred conversations a month will not carry fifty thousand, and nobody planned for that. This is the path we use, and what has to be true before each step.

3
months to the next process
4
steps, each with a checklist
16+
languages, when you need them

Why the second process is harder than the first

A pilot is allowed to be fragile. It runs in one channel, in one language, for a handful of people who forgive it. Growth removes all three of those cushions at once: more channels, more languages, more people, and colleagues who did not ask for any of it. The technical part is rarely what stops it. What stops it is that nobody agreed who owns the change, who pays for the next step, and what would count as it having worked.

Four steps

One process, proven

Enough to find out whether the idea holds, and cheap enough that finding out costs little.

What it covers

  • One or two channels
  • One language
  • A hundred to a thousand people
  • Connected to whatever is easiest

What it runs on

  • Shared servers
  • Basic monitoring
  • Working 99.5% of the time
  • Support during business hours

How you know it worked

  • It answers correctly more than 85% of the time
  • The people using it rate it above 4 out of 5
  • The time saved is measured, not estimated
  • The case for spending more is now on paper

One department

The whole team works this way, and it is connected to the systems the company actually runs on.

What it covers

  • Three to ten channels
  • Two to four languages
  • A thousand to fifty thousand people
  • Connected to your CRM and back office

What it runs on

  • Your own dedicated setup
  • Fuller monitoring
  • Working 99.9% of the time
  • Extended support hours

How you know it worked

  • The money it returns is measured, not projected
  • More than 80% of the team uses it by choice
  • The process itself got simpler, not just faster
  • The cost per case went down and you can show it

The whole company

Every department that needs it has it, and it is treated as part of how the company runs.

What it covers

  • Ten to fifty channels
  • Eight or more languages
  • Fifty thousand to half a million people
  • Connected to everything that matters

What it runs on

  • Your own servers, or a separate one of ours
  • Detailed monitoring
  • Working 99.95% of the time
  • Support around the clock

How you know it worked

  • It appears in the company's own plans, not ours
  • Someone on the board owns it
  • Competitors notice
  • It is now something you sell against

Several countries

The same thing running in more than one market, close enough to each to feel local.

What it covers

  • Fifty to a hundred channels and more
  • Sixteen or more languages
  • Half a million people and up
  • Several companies or brands on one system, with the data kept apart

What it runs on

  • Servers in more than one region
  • Fast everywhere, not just near you
  • Working 99.99% of the time
  • A named team on your side

How you know it worked

  • You are ahead in your market, not catching up
  • The system itself is part of what you sell
  • Partners build on top of it
  • It brings in money rather than only saving it

What changes by industry

The path is the same; what gets connected and what has to be proven is not. These are the ones we have done more than once.

Property

  • Sorting inquiries at volume
  • Several buildings or portfolios at once
  • Connected to thirty-plus property systems
  • More than one currency and language

Rental

  • Connected to your fleet records
  • Prices that move with demand
  • Several locations kept in step
  • Connected to payment providers

Crypto and finance

  • Support that does not sleep
  • Built to the rules of your regulator
  • More than one network supported
  • Warnings when something looks wrong

Online retail

  • Connected to your catalog
  • Order status answered automatically
  • Returns and refunds handled
  • Stock kept in step

Healthcare

  • Built to the rules for medical data
  • Appointments booked and moved
  • First contact with the patient handled
  • Connected to the patient record

Banking

  • Built to the rules for card payments
  • Connected to your core banking system
  • Support for spotting fraud
  • Regulatory reporting prepared

Before you take the next step

On the technical side

  • You know how much your own systems can take before they slow down
  • There is a decision about how the data store handles the growth
  • Pages and answers come back fast, wherever the person is
  • It has been tested at ten times the load it carries now
  • There is a plan for a total failure, and it has been rehearsed
  • Alerts are set up and someone receives them

On the human side

  • One person at the top owns this and says so out loud
  • There is a plan for how people are told and trained
  • The training material exists
  • It is clear who gets called when something breaks, and after them who
  • Everyone has agreed what would count as this having worked
  • The money for the next step is approved, not hoped for

How long each step takes

  • One process, proven4–6 weeks
  • One department6–8 weeks
  • The whole company8–12 weeks
  • Several countriesongoing

Related

Next step

Measure it before you build it

Everything above rests on one thing: the process needs a measured baseline before anything is built on top of it. Thirty minutes to see whether the arithmetic is likely to close, and the sprint that follows comes back in full if we conclude you should not build.

Put a time in the calendar

Thirty minutes, free. The sprint is what the call is about.

Fee
$2,500
Length
1-2 weeks

Refunded in full if we conclude you should not build.

Or start with the free 15-minute diagnostic

Updated: August 2026