CloseMenu

02 / Systems Sprint

Ops that stopped retyping

When the tools do not talk to each other the people become the integration, and the first two hours of the day go on carrying the same facts between systems by hand. Cut the retype path below and the first two hours come back as a morning that just runs.

  • 5 sample mornings
  • Hours back, weekly
  • 6 weeks
  • NL / EU

Before you scrub

What the scrub is showing you

  1. Retyping is the bug

    If somebody copies the same field between two tools every morning, that path is broken, and it stays broken however good the people doing it are.

  2. Automation needs a trail

    A workflow with no log behind it, no gate in it and nothing to undo it is a workflow your team will quietly work around within a fortnight.

  3. The hours are the tell

    Wherever the morning burns hottest is where the retyping is worst, and that loop is the first one I put into production.

Jump to Morning Cut ↓

Morning Cut

Scrub the morning. Cut the retype.

Five sample loops, built from the work I am most often shown when somebody says the mornings have stopped working. Drag the clock while the hours burn, then cut the path and watch them come back. Whichever loop runs highest in your business is usually the first one I ship.

Broken Monday · Line 01

Invoice match

An invoice PDF arrives by email and its line items land in the system that runs your accounts, the ERP, without anyone retyping them.

Finance teams drowning in email attachments.

Example figures, not a client’s.

Burning this morning1.7hPeak 6h / week if nobody cuts
09:34Scrub the morning
  1. 01Inbox
  2. 02PDF
  3. 03Sheet
  4. 04ERP

Queue

  • INV-1842 · AcmeStuck in sheet

Clipboard still moving — each hop is a person retyping.

After the cut

Start from your worst loop.

Every Systems Sprint starts from whatever your team retypes most. I diagnose where the operating day actually goes, we agree a build order between us, and one or two paths then go into production with the trails and the human gates that let people keep running them once I have gone.

The five steps

How a Systems Sprint runs

6 weeks, in the 5 steps below, and the last of them is the only test of this work that actually counts.

  1. Diagnose

    I map the loops burning the hours, measuring volume, friction, and who currently owns each hop in the chain.

  2. Build order

    The one that gives back the most time first, written on a single page that both of us agree before anything gets built.

  3. Wire

    The systems that hold the real records get wired to each other, and the path that made somebody retype the same field twice disappears.

  4. Trail + gates

    Logs somebody can open on a Thursday and follow without ringing you, with people still deciding on the exceptions and anywhere the stakes run high.

  5. Handover

    Your team runs it after I leave. That is the only test of this work I have ever found worth applying.

FAQ

Questions about Systems Sprints

What is Morning Cut?

An interactive scrub of a broken working day. You drag the clock forward while the retyping burns hours off the morning, then cut the path and watch the hops wire themselves together, a trail appear, and the hours come back. It illustrates how a Systems Sprint feels from the inside; it is nobody's live dashboard.

Is this only the workflows on the board?

No. The five mornings are the ones I am asked about most often, and any sprint starts from your highest-volume retype loop, whether that sits in finance, sales, support, reporting, onboarding, or somewhere none of those describes.

Do you drop AI into every path?

Only where it earns its keep against the manual path it replaces. A good share of the wins here are plain automation plus one agreed place where the real number lives. Deciding where a model genuinely belongs is most of the work, and it is the part people skip.

What happens when an automation breaks?

It stops somewhere you can see, which is what the trail and the gates are for. A workflow that fails loudly on a Tuesday morning costs you an hour; the same workflow failing quietly costs a month, because by the time anyone notices the bad data has spread into everything downstream.

How do you handle compliance and legal advice?

I ship systems that keep a log, hand off to a person when they should, and stop where a human still decides. Where the question is really a legal one I say so and we bring in a lawyer, which covers AVG and AI Act opinions: I am not your lawyer, and this practice has no Data Protection Officer.

How long is a Systems Sprint?

6 weeks, covering the diagnosis and 1 or 2 live workflows, agreed in writing before the first thing is wired. Handing it over sits inside those 6 weeks, because a sprint that ends with your team unable to run the thing has not finished.

If that Monday looked like one of yours, that is the point of showing it. The client stays unnamed; the hours are what you would actually get back.

Last reviewed