What the layer does once it is actually running.
Three companies, three industries, three very different settings for how much authority they gave the layer. Each one started with a single department.
11
Departments running on the layer
4.2M
Decisions routed or executed
68%
Decisions cleared without a meeting
19 days
Median time to first autonomous tier
Northbeam Supply
Margin decisions that used to take a week now take an hour.
Industry
Industrial distribution
Size
740 employees
Region
Rotterdam, Netherlands
Departments
Finance, Operations, Sales
The challenge
Northbeam distributes industrial fasteners across fourteen European markets, with pricing set regionally and inventory held centrally. When a supplier raised prices, the change reached the sales team through a spreadsheet that was already three weeks old.
By the time anyone reconciled the new landed cost against committed quotes, the company had usually sold several thousand units below margin floor. Nobody was at fault. The information simply could not travel between finance, operations and sales faster than the deals closed.
Their existing automation handled invoice capture well. It could not tell anyone what the invoice meant for the quote a salesperson was about to send.
What they deployed
Northbeam connected their ERP, quoting tool and supplier contracts to one knowledge base, then put finance and operations on the intelligence layer first. Sales joined six weeks later once the margin logic had been validated against a quarter of historical deals.
One pass of the loop
Update
A supplier invoice lands showing a 9% increase on a fastener line held in three warehouses.
Monitor
The layer finds 214 open quotes and two framework agreements that price against that line.
Decide
It works out which quotes now fall below the 12% margin floor, and what each would cost if honoured.
Approve
Repricing inside the agreed band clears automatically. The two framework agreements route to the commercial director.
Act
Quotes are updated in the quoting tool, and the affected reps are notified with the reasoning attached.
Improve
Win rates on repriced quotes feed back, and the layer tightens its estimate of which increases customers will absorb.
6 days
Cut from cost-change to quote update
€1.4M
Below-floor revenue avoided in year one
91%
Repricing decisions cleared autonomously
We had automated the paperwork years ago. What we had never automated was the judgement in between, and that was the part actually costing us money. The layer does not just tell us a cost went up. It tells us which eleven quotes are now wrong and fixes the ones we have agreed it can.
Mirjam de Vries
Chief Financial Officer, Northbeam Supply
Calyx Health
Support signal reaches the roadmap the same day it appears.
Industry
Digital health services
Size
310 employees
Region
Austin, United States
Departments
CS/CX, Product, Marketing
The challenge
Calyx runs a subscription care platform with a support team handling around 9,000 conversations a month. Patterns in those conversations were the single best predictor of churn, and almost none of it reached the product team in a usable form.
Support tagged tickets diligently. Product read a summary once a fortnight. By the time a defect was prioritised, it had usually been costing renewals for two months.
Because Calyx operates in a regulated space, they were also clear that no decision touching a patient account could execute without a named human approving it.
What they deployed
Calyx put CS/CX and product on the layer with every decision class capped at Tier 2, so the platform prepares the action in full and waits for approval. Marketing joined later to close the loop between messaging and support volume.
One pass of the loop
Update
Conversations mentioning a failed pharmacy sync rise sharply over four days.
Monitor
The layer clusters them, links the accounts to revenue, and matches them to an open engineering ticket.
Decide
It ranks the defect against current roadmap items by revenue at risk and renewal proximity.
Approve
The reprioritisation and a proactive customer message are prepared and sent to the head of product to approve.
Act
On approval the ticket is reprioritised, affected accounts are messaged, and support gets a briefed macro.
Improve
Contact rate after the message is measured, and the layer learns which proactive notices actually reduce volume.
4 hours
From support signal to prioritised defect
31%
Reduction in repeat contacts on known issues
100%
Patient-facing actions human-approved
The tiering is what let us say yes. We were never going to hand a system the ability to message patients on its own, and we did not have to. It does all the work and then stops and asks. That is the right division of labour.
Daniel Okonkwo
VP Customer Experience, Calyx Health
Meridian Freight
The capacity desk now runs itself, and reports like a manager.
Industry
Logistics and freight forwarding
Size
1,250 employees
Region
Singapore
Departments
Operations, Finance, Sales, CS/CX
The challenge
Meridian books ocean and air freight across Asia-Pacific, where committed capacity, spot rates and customer commitments move daily. Their capacity desk was four people reconciling three systems and a shared inbox.
The work was pure coordination: match committed demand against held capacity, spot the shortfall early enough that it was still cheap to fix, and tell sales before they promised something the company could not deliver.
The team was good at it, and it did not scale. Every new trade lane meant another person, and mistakes were concentrated in the hours nobody was awake.
What they deployed
Meridian started at Tier 1 for a single trade lane, moved to Tier 3 after a quarter, and now runs the capacity desk as a Tier 4 pilot where the layer owns the function outright and reports on it weekly to the operations director.
One pass of the loop
Update
A carrier reduces allocated space on a Singapore to Rotterdam lane by 18% for the coming month.
Monitor
The layer reconciles the new allocation against committed bookings and the current sales pipeline.
Decide
It identifies a 240 TEU shortfall, prices three alternatives and picks the one that protects contracted customers first.
Approve
Within its mandate it proceeds. The spend sits inside the ceiling the operations director set.
Act
Space is secured with a secondary carrier, bookings are reallocated, and sales sees the revised lane capacity.
Improve
Realised cost against forecast is written back, sharpening the next allocation call on that lane.
3x
Trade lanes covered without added headcount
22%
Reduction in emergency spot bookings
< 15 min
Median response to a capacity change
I review its week the way I review anyone else on my team. It brings me the decisions it made, what they cost, and what it got wrong. Honestly, the write-up is better than most of the ones I used to get.
Priya Raghunathan
Director of Operations, Meridian Freight
Every one of these started with one department.
Bring the one where decisions wait longest. We will show you the loop running on your own data before you commit to anything.
Currently onboarding design partners.