/ Portland, 3:02 PM
Support specialistSupport specialistOpen a live threadwith the studio desk
(NSP® / 01)Selected work

Systems we designed, built, and still run inside operations that cannot afford a bad week.

Case Studies

Calderwood FreightCalderwood

Operational mapping

Calderwood Freight

We rebuilt the freight planning loop and shipped systems that schedule routes, capacity, and delivery windows without a planner in the middle.

Year
2025
Industry
Logistics and supply chain
Work delivered
Operational mapping, automation build.
Challenge
Planning slowed as shipment volume grew
Stack we use
Deliverable: a deployed planning system
Read the case
Halden Retail GroupHalden

Operational mapping

Halden Retail Group

We deployed decision systems that hold inventory steady across sites and take the guesswork out of weekly replenishment.

Year
2025
Industry
Retail and commerce
Work delivered
Operational mapping, bespoke AI systems
Challenge
Manual planning caused stock imbalance
Stack we use
Deliverable: a live replenishment engine
Read the case
Vantage Mobility NetworkVantage

Automation build

Vantage Mobility Network

We automated dispatch decisions and gave controllers one live view of a fleet that used to live in four spreadsheets.

Year
2025
Industry
Mobility and transport
Work delivered
Automation build, system operations
Challenge
Dispatch relied on tribal knowledge
Stack we use
Deliverable: one dispatch control surface
Read the case
Solstice EnergySolstice

Data infrastructure

Solstice Energy

We built the data layer and monitoring that lets an energy operator act on faults in minutes instead of shifts.

Year
2026
Industry
Energy and utilities
Work delivered
Data infrastructure, system operations
Challenge
Fault data arrived too late to act on
Stack we use
Deliverable: a fault-detection data layer
Read the case
Abstract dark silk render
(NSP® / 13)Questions we get

The Short
Answers

The questions operators actually ask us, answered here so nobody has to sit through a call for them.

Ask the studio
1/4

No. A general model is one component among several, and usually the least interesting one. The work is in the retrieval over your own records, the guardrails, and the logic that decides what happens next. Swap the model out in a year and the system should barely notice.

Most teams see a measurable change inside the first quarter. Removing repeat work or shortening a decision shows up in cost and capacity long before it shows up on a slide.

Yes, and that is usually the cheap part. We connect through whatever you already expose: a service endpoint, a read replica, a message queue, even an overnight file drop. The system lands inside the current workflow rather than beside it.

Security is a design constraint here, not a review at the end. Processing stays isolated, storage stays encrypted, access is scoped per role, and nothing leaves the boundary you set on day one.

Most questions come from operators
(NSP® / 14)Reach the studio

Start here

Bring a system, a workflow, or a rough idea on a napkin. We will give you the honest shape of building it and keeping it alive.

+1 (555) 218-0140
hello@nullspace.example
PIER 9, STUDIO 400, PORTLANDREMOTE BY DEFAULT, CLIENTS WORLDWIDE

Find us on

What are you looking for?
Current monthly AI spend

By submitting you agree to our demo privacy note

Demo form. Nothing was sent, and no data left this page.