Services
What we do for you as an engagement: six service lines, an independent review of someone else’s system, and optimisation and sensing work on the engine we run. Every one is scoped and quoted in writing before it starts.
Six service lines
Each line has its own page: the problem it starts from, what the engagement leaves behind, and its honest limit.
Marketing measurement
Know which spend actually moved the number.
Decision systems
Turn a recurring judgement call into a system that explains itself.
Scheduling & logistics
Rosters, routes and capacity that hold up on contact.
Industrial vision
Count, track and flag what the cameras already see.
Data & ontology
One vocabulary your systems and your people both use.
Edge & hardware
When the decision has to run on the machine itself.
A review of someone else’s system
A seventh service, kept out of the Services menu: a review of an AI or system you did not build, for whoever is buying, funding or insuring it.
Every engagement has the same three shapes
- Diagnostic. About a week, for a fixed fee. It ends in a written assessment, and it is allowed to end there.
- Pilot. A few weeks, for a fixed scope and a fixed price, judged by a success measure we agree in writing first.
- Build and hand over. Capped time and materials, or a fixed price per milestone, and your team owns what is built.
We publish no figure for any of the three: each is priced in a written quote for your scope. How we work describes each shape in full.
Optimisation engagements
Six engagements, each its own scoped piece of work:
Logistics Optimization
Supply chain, routing and capacity, turned into an auditable Pareto frontier — the whole
trade-off surface, not one answer, so the trade-offs are explicit rather than buried in a
spreadsheet. The frontier step is the one embark.day serves as
decisions.frontier: send
your options and your objectives, and get every option that no other option beats on all of them at
once.
Process Integration
Day's optimisation kernels wired directly into your existing workflows.
Agile Delivery
Sprint and capacity planning treated as a constraints problem first, so the plan holds up against the constraints you actually have.
IT Consulting
Digital transformation built on the Day substrate.
Build a Kernel
Ship a domain-specific optimiser as open source.
Launch a Product
Use the optimiser as your engine, and build your own product's domain layer on top of it.
Every one of the six is dispatched through the same engine: whatever the problem, it is formulated as a QUBO and solved at whichever tier — exact or by simulated annealing — fits its size and the latency you need.
Environmental sensing, on the register
Where to put a sensor, and what the forecast at that spot is really going to do, is the same kind of work as the six above: a coverage problem under a budget, and a correction learned from measurements you already collect. It is on the register as an engagement, and it is also a product you can use without hiring anyone.
Use it yourself. Weather & sensing is the live version — a public forecast, the reading from the station inside the same grid cell, and the measured gap between them at every monitored point, each labelled with whoever published it. Your own weather stations is the half you build with: connect the station you already own and the forecast at your coordinate is corrected against your readings, with the fit reported rather than implied.
Hire us for it. A network larger than the free demo, a variable nobody has scored at your sites yet, or a correction that has to feed an existing operational decision is an engagement — scoped, fixed-fee, and allowed to end at the diagnostic if the honest answer is that the public forecast is already good enough for you.
What is measured, and where it stops. On the curated demo network the placement answer covered 14 of 16 demand units at a budget of 4 sites from 8 candidates, and it names the two regions it leaves uncovered rather than reporting coverage alone (re-solved 2026-09-17). The correction's skill against the raw public field is published per study site on the product page, read live from the service, with the site where it barely helped shown next to the sites where it did. A live sensor beats the corrected forecast from a lead of roughly one to two hours in summer and up to six in winter — which is the argument for putting one there, not for buying a forecast from us. We are not a forecast authority and not a warning service.
Where an engagement starts
With a short conversation about one decision. If we can help, the next thing you receive is a written scope for a diagnostic or a pilot, with its price. If what you want is something your own team runs, the products are the other half of what we sell.