Product tour

See your plant move. Test the plan. Then commit.

Planning software doesn’t have to look like a spreadsheet from 1998. Unbrick shows your shop as it is, lets you try changes on the whole plan before anyone touches the floor, and keeps every number tied to the record behind it.

The plant map

Your plant, moving

Press play. The last eight weeks replay from your labor records, then the plan carries every lot forward through your departments. In 3D you see the machines and the people too. Each lot is a block; click a department for the lots, machines and people behind it.

The plant map in 3D on a screen by the floor: each department with its machines and the lots on them or queued in front, and two flagged constraints, deburr and inspection, each with the reason, such as inspection waiting on the two people certified to program the CMM
Demo Precision Works, a synthetic plant. Red blocks are lots late against the plan; machines light up as they get busy.

History, then the plan

Where every lot was each day, from labor records, and where the plan takes it next. Scrub to any day, play it, or walk the camera through the plant in flow order.

The constraint, flagged

Each day, the departments with the most work waiting while they are busy, marked with the reason: people, a scarce skill, or machines, and since when.

Your floor, on the screen

Drag departments over a picture of your floor plan. Put it on the TV by the floor: four weeks back, today, four weeks ahead, on a loop.

People & skills

Plan on the people who can do the work

A department is not one pool of hours. Two people may hold the precision grinding, and one may run two machines at once. Unbrick plans on both: the work goes to people who hold the skill, on the machines that run it, and a person running two machines is counted once.

  • Skill levels proposed from each person’s own labor records, with the records behind them
  • Confirmed by a lead or planner; certificates that expire stop counting
  • Skills at risk, absences and a training plan in one place
  • Each constraint says what binds it: “320 std h waited for the 2 people certified to program the CMM”
The skill matrix filtered to skills at risk: deburr people with levels proposed from their records, and CMM programming marked as needing a certificate, with two certified programmers counted, an expired certificate hatched, and a trainee and an unconfirmed proposal faded because they do not count
Synthetic demo plant; names are invented.
Cell boards

Give each cell its own board, and the say over its queue

The lead of each cell sees the work in front of it against its people and machines, who the plan gives each job to, and how much of last week’s plan got done. They can move a job up, hold it, or give it to someone, with a reason. Each change goes into the next plan, which says whether it was applied or why not.

Owned, and on record. A weekly commitment and review per cell, kept as written. Hot work and plant holds still come first.

The grinding cell board: lots and days of work in front of the cell, last week’s plan attainment, what limits it, the load by day against people and machines, and the cell’s weekly commitment
Synthetic demo plant.
Side by side

Two plans, playing in step

Put today’s plan beside a staffing scenario or an engine’s proposal and play both. Badges on each department show where fewer lots wait. It’s the quickest way to show a plant manager what a change actually does.

  • Today’s plan, a saved scenario, or any engine’s proposal
  • The same days and the same history on both sides
  • Nothing changes until a planner adopts it, with a reason
Two plant maps side by side: today’s plan and a crew allocation proposal, with badges showing ten fewer lots waiting at inspection
Synthetic demo plant.
Order journey

“Why is this order late?” in one picture

Click any order line. You see every lot behind it, the components and stock that feed those lots, what each one has done and what the plan says comes next. The lots that set the ship date are marked, and the reason is spelled out in plain words.

Then ask. One click opens the assistant with the question filled in. It answers from the same records and cites them.

An order journey: the assembly lot and its component lots on a timeline, worked steps solid, planned steps outlined, with the need date and today marked
Synthetic demo plant; customer names are invented.
Plan engines

Solvers that show their work

Start with a rule your planners can predict. Add solvers as the shop grows. Every engine plugs into the same plan, and none of them gets the last word.

1

Propose

An engine suggests an order of work, or which cross-trained people to move, and says what it aimed for and what limited it.

2

Rerun the whole plan

The proposal is replayed through the same finite plan as everything else. No solver is taken at its word.

3

Score it

Order lines on time and late, days late, what moved inside the frozen window, and why each lot moved.

4

A planner decides

Adopt it with a reason, on record, or don’t. If a solver ever fails, the plan falls back to the simple rule and says so.

The engines today

From a rule anyone can follow to mixed-integer solvers

Due date first

The default. Hot, then expedite, then the earliest need date.

Least slack

A published dispatching rule: what’s at stake per day of work, discounted by slack.

Bottleneck sequencer

Google OR-Tools CP-SAT orders the work at your top constraint to cut weighted lateness.

Crew allocation

HiGHS finds which cross-trained people to move between departments, and for how long.

Open-source solvers (Apache-2.0 and MIT), running on ordinary CPUs inside your deployment. Your data goes to no solver service.

A crew allocation proposal: lines on time, lines late and total days late before and after, how it decided, and the solver report
Synthetic demo plant. The figures describe the simulator, not a customer.
Sequence board

Rearrange the bottleneck, see what it does

The order work runs at one department over the next weeks. Drag lots to try another order. Work inside the frozen window is locked, because the floor is already on it.

Preview reruns the whole plan and shows whose orders gain and whose slip. A planner can make it the order, with a reason.

The sequence board for one department: locked lots inside the frozen window, then lots that can be moved, with their planned bars and need dates
Synthetic demo plant.
Every day

The rest of the toolkit

Flow

Where work goes next, department by department, in the plan or over recent weeks. The thick bands show where the hours pile up.

Load by week

Every department, every week, against its capacity. Click a week to open the sequence board.

Staffing scenarios

Add people, move them, add a shift or overtime, and see when past due recovers. Capacity splits into people, hours and efficiency, from your labor records.

Throughput

What can ship today, what’s nearly done for late orders, what has stalled, and where one setup could serve several lots.

Planned orders

The new lots it would take to cover demand nothing covers, and “when could we have 50 more?”, answered from what the plan leaves free.

Actions

One list each morning: ship, restart, review a hold, release, move an ERP date. Anything inside the frozen window is marked.

The flow view: departments in flow order with bands showing standard hours moving from each department to the next
Synthetic demo plant.

Every dot opens a record

Lots, labor, routings, orders. Nothing on a picture is there for decoration, and nothing is estimated without saying so.

Drawn in your browser

No map, chart or font service is called, hosted or on your own server. The 3D view needs a browser with WebGL; without it, the flat and iso views still work.

Measured on your data

The engine backtest replans from your earlier days and compares deliveries, date error and day-to-day churn. We don’t quote demo results as yours.

See it with your own eyes

A 30-minute demo on the synthetic plant

We play the plant map, follow a late order to its cause, run an engine and adopt it, then show you how the same thing runs on your data in a 30-day pilot.

  1. WatchYour plant moving: history, then the plan.
  2. Ask whyOne late order, traced to the lot that sets its date.
  3. Try a changeA solver proposal, scored on the whole plan.