Skip to content
Paid pilots open

Root cause analysis that proves the fix held.

Cadence Resolve coaches warehouse and logistics teams through a real investigation, holds every corrective action to an owner and a date, and then asks whether the problem actually stopped. The loop closes on evidence, not on a ticket being marked done.

  • 300+ automated tests
  • Tenant isolated data
  • No training on your data
  • UK built and run
An incident branching into hypotheses, two of which become verified system causes, each with a corrective action, both feeding one effectiveness check that points back at the incident.INCIDENTHYPOTHESISHYPOTHESISHYPOTHESISVERIFIED CAUSEVERIFIED CAUSECORRECTIVE ACTIONCORRECTIVE ACTIONEFFECTIVENESS CHECK

Incident, hypotheses, verified causes, actions, effectiveness check.

How one incident becomes a verified fix
StageIn this exampleLeads to
IncidentIncident: repeat mispicks on the night shiftHypothesis: picker was careless; Hypothesis: look-alike stock sits together; Hypothesis: no bin check at the scan
HypothesisHypothesis: picker was careless (discarded, the evidence did not support it)The case closes, and the check reopens it if the problem returns.
HypothesisHypothesis: look-alike stock sits togetherVerified cause: slotting puts look-alike SKUs side by side
HypothesisHypothesis: no bin check at the scanVerified cause: the system asks for no confirmation
Verified causeVerified cause: slotting puts look-alike SKUs side by sideAction: reslot the fast movers, owner and date set
Verified causeVerified cause: the system asks for no confirmationAction: scan to confirm the bin on look-alikes
Corrective actionAction: reslot the fast movers, owner and date setEffectiveness check: did it hold?
Corrective actionAction: scan to confirm the bin on look-alikesEffectiveness check: did it hold?
Effectiveness checkEffectiveness check: did it hold?The case closes, and the check reopens it if the problem returns.
01The problem

Closing the ticket is not the same as fixing the problem.

How it usually goes

  • The loudest theory in the room wins, and it is usually a person.
  • The action list says "retrain the team" with no name and no date against it.
  • The investigation lives in a spreadsheet nobody opens again.
  • Nobody checks in six weeks, so the same failure comes back next quarter and is treated as new.

What we hold you to

  • A cause you can change: slotting, a system prompt, a standard, a layout.
  • An action with an owner, a due date and a date to check whether it worked.
  • A close gate that refuses a case missing any of those.
  • A verdict on the check date, and if it recurred, the case reopens as an 8D with the history attached.
02How it works

Walk through one investigation, start to signed A3.

Describe the incident

One sentence is enough to start. The coach picks a method and tells you why it picked it, and you can override it.

YouThird mispick this month on the aisle 7 fast movers, night shift.

CoachStarting a 5-Why: one incident, one chain of causes. Switch to a fishbone if more than one factor appears.

Illustration of the coaching pattern. The example is fictional.

03Methods

One investigation, whichever method it turns out to need.

The method is not a menu you pick from at the start and regret later. It follows the evidence, the coach suggests a change when the shape of the problem changes, and a person makes the call.

One incident with one chain behind it.

5-Why

The chain persists as a tree, so the fifth why is attached to the fourth rather than living in someone's notebook.

More than one factor is in play.

Fishbone

Causes are grouped by People, Process, Equipment, Layout and slotting, WMS and system, and Environment. The coach suggests the switch when a second category appears.

Someone above you needs to read it.

A3

Not a separate exercise: the A3 assembles itself from the investigation you already did.

It has happened before, or it is serious.

8D

The eight disciplines in order, with interim containment as a gate. A case cannot close until containment is recorded.

04The artifact

The A3 is the point. Everything else assembles it.

An A3 is what survives the meeting. This one is generated by the product from a worked example, using the same code that prints yours: the why-chain tree, the fishbone, the actions with their verdicts and a sign-off block.

The example uses a fictional site and a fictional incident. It is not a customer's investigation.

05The pilot

A paid 30 day pilot on one site.

£1,500 to £3,000one site, 30 days, invoiced once

  • One site, up to five investigators, three real investigations run end to end.
  • Your workspace, isolated from every other tenant, with sign-in locked to your work email domain.
  • A working session to set it up, then coaching by email while your team uses it.
  • The A3, fishbone and change-plan prints for every case you run, yours to keep.

What we need from you

  • One named owner on your side who can pick the three investigations.
  • Investigators who will type what actually happened, not a tidy version of it.
  • Half an hour at the end to tell us what was useful and what was not.

What happens after the 30 days

  • If it earns its place, the ongoing price is £500 to £1,000 per site per month.
  • If it does not, you owe nothing further. We export your data and delete the workspace.
  • Either way the prints you produced during the pilot are yours.
06Security and data

Your investigations are your data.

Isolation

One tenant per company

Every query is scoped to your tenant, and cross-tenant isolation is covered by the automated test suite that runs on every change.

AI

No training on your data

Coaching turns go to Anthropic's API under commercial terms that exclude training on API data. Nothing you write trains a model.

Access

Sign-in tied to your domain

A single-use link to your work email, and during a pilot we lock sign-up to your domain so nobody else can reach the workspace.

Exit

Export, then purge

Ask, and we export the whole tenant and delete it. Coaching transcripts on closed cases age out automatically after a year.

Read the full security note or the privacy notice.

07Questions

The things buyers actually ask.

How much work is this for my team?
One setup session, then the tool is used inside investigations you were going to run anyway. There is no data migration, no integration and nothing to install. Investigators sign in with a link sent to their work email.
What happens to what we type into it?
Incident descriptions, causes, actions and coaching conversations are stored in your own tenant and are only visible to people signing in from your work email domain. We do not sell or share your data.
Does the AI decide anything?
No. It asks the next question, proposes causes and recommends a method with its reasoning. A person records the cause, changes the method, records the effectiveness verdict and closes the case. The software will not do any of those on its own.
Is our data used to train models?
No. Coaching turns are sent to Anthropic's API under commercial terms that exclude training on API data. Nothing is used to train a model, ours or anyone else's.
Who can see our investigations?
Everyone in your tenant sees the tenant's cases. There are no per-user roles yet, so a pilot team should be people who would sit in the same investigation anyway. On our side, access is limited to the founder, who holds the infrastructure credentials.
Can we get our data out?
Yes. Every case prints as an A3, a fishbone and a change plan from the product itself. On request we export the whole tenant and then delete it.
What if we want to stop?
Tell us. The pilot is a fixed 30 days with no automatic renewal, and there is no notice period to serve. We export your data and purge the workspace.
Why is the pilot paid?
Because anything given away gets treated as such, by both sides. Paying for the pilot buys real attention: your team commits three real investigations, and we commit the setup and the coaching. The price is set to be a line item rather than a procurement exercise.
08Next

If one problem keeps coming back, start with that one.

Applications take about five minutes and go straight to us. We take a small number of pilots at a time, so tell us the problem you would point it at. After the pilot the ongoing price is £500 to £1,000 per site per month.