The Axiovex method

Constraint-based planning.

A constraint is a limit that is actually true: an opening that really exists, a course that really runs, a credential that takes as long as it takes, a lab with only so many stations, a person who can be in only one place at a time. Constraint-based planning means collecting those limits first — all of them — and building only the plans that satisfy every one at the same time.

What it means

The limits come first. Then the plan.

The limits we plan inside include: openings that are confirmed, not forecast; the skills a job actually requires; the courses and credentials that build those skills, and how long they take; class and cohort schedules; the places involved and the distance between them; who is available, and when; and the capacity of the rooms, labs, machines and people everything depends on. A plan earns the name only when it satisfies all of them at once.

That sounds obvious. It is not how most plans are made. Most plans start from an average, a target, or last year’s pattern, and discover the real limits later — one collision at a time, usually after commitments have been made on the strength of them.

This is also the method behind our flagship project: a data pipeline we are building — Michigan data in; insights and recommendations out, for employers and programs, and for the Michigan economy as a whole. Its first public outputs are already live — Axiovex Signals, our daily board; The Axiovex Signal, our monthly newsletter; and the monthly Michigan Workforce Intelligence report — and every recommendation the pipeline produces is built this way: from the real limits, all at once. Axiovex works both sides of that: consultants and researchers today, and the builders and future managers of the program the research adds up to: Axiovex Arcturus, a workforce and capability program for participating employers, universities, and workforce organizations. Its planning system uses a projected constraint system — the method on this page, carried into the program’s operations and fed by the pipeline. Bringing clarity, and building a program that turns that clarity into workable plans, is what the company is for.

A worked example — illustrative

One cohort of welders

Say an employer needs welders starting in June, and the credential takes a full term. A plan built on the average would enroll a cohort in January and hope. A constraint-based plan checks the real limits: the course sequence and its actual length, the lab’s stations, the instructor’s hours, the start date that cannot move, and each participant’s work schedule and drive.

If the cohort would finish in July, the conflict shows up now — and the plan changes while changing is cheap: an earlier start, a split cohort, or a second start date the employer agrees to. And if none of those work — if the time simply is not there — the method says so, and names time as the binding constraint, instead of letting June arrive with no welders.

Why it’s powerful

Four things this discipline buys you.

Feasible under the model — or honestly reported

Every plan this method produces is checked against all the documented limits at once. And when no plan fits, it says so — and names the constraints that bind, the limits actually in the way. Most failed plans were never possible; they were just never checked. Checking is the whole job. A check ends in one of five reported states, and the state is part of the answer: feasible under the model; proven infeasible under the model; unknown, or timed out before a proof either way; incomplete or stale inputs, where the documented limits cannot support a verdict; or held for human review. Feasible under the model is a statement about the model and the inputs as documented and dated — it is not a guarantee about the world.

Conflicts surface before commitments

A course that finishes after the job starts. A cohort larger than the room it trains in. Two employers counting on the same graduate. Every conflict found on paper costs a conversation. Every conflict found in June costs a cohort, a contract, or a reputation.

Trade-offs in the open

Real limits force choices: the same people, rooms and hours cannot serve two plans at once. The method makes those choices explicit — what was traded, for what, and who decided. A plan that hides its trade-offs doesn’t avoid them; it just decides them by accident, and nobody can revisit a decision nobody knows was made.

Re-solved when the facts change

Openings close. Schedules move. People drop out, and new people arrive. When a fact changes, the plans it touches are re-checked against the limits. The world does not hold still for a plan — re-checking is the difference between a plan that degrades quietly and one that stays true.

What we don’t sell

We don’t sell seats — that is a college’s work. We don’t sell software — that is a vendor’s. We don’t sell staffing — that is an agency’s. What Axiovex provides is the plan itself: the plan that fits your documented constraints, found and checked. If a seat, a software license, or a staffing contract is what you need, buy it from the people who sell those well. If what you need is to know whether the whole thing fits together — and what to change when it doesn’t — that’s the work we do.

One method, across the firm

The limits differ. The discipline doesn’t.

The pilot

The workforce pilot and employer planning

One institution, 10–20 employers and 20–30 participants plan around a single occupation family, inside real openings, course schedules and capacity. The pilot is Arcturus’s first cohort — the method’s first full test, run in the open and measured.

The data

Signals and the monthly workforce report

Signals and our monthly Michigan workforce report keep the facts current and honest — published, dated and sourced. This is why we publish the data instead of just using it: constraints you can inspect are constraints you can trust.

AI work

AI readiness and adoption

Start from a company’s real limits — its people, its data, its time, what it can afford to risk — and find the adoption plan that fits. Most AI plans fail on limits nobody wrote down: data that isn’t ready, people who weren’t consulted, time that wasn’t there.

The build

How we build systems

Requirements first, limits first, then build — and the build has to satisfy them. A build that can’t meet its requirements is stopped by its own plan, not by its launch date. Solving problems and building things run on the same rule.

Where this leads

Program scale, measured on real jobs.

The goal is to put constraint-based planning to work at program scale in Michigan: to plan workforce programs this way — and, in time, to operate them — and to judge them the way the method judges any plan, on whether they fit reality. For a workforce program, that means one measure above the rest: whether people land in real jobs, and stay in them.

No single party holds all the limits — not one employer, one school, or one agency — so the work has to connect across the wider ecosystem: employers, associations, workforce boards, educators, and training operators. And it has to keep its facts public: the monthly workforce report and Signals are how we prove, in the open, that the picture the plans are built on is true.

Clear edges

What the method is not.

Not a forecast

It does not predict openings or enrollments. Forecasts stay labeled as forecasts; plans are built on confirmed facts.

Not planning by averages

An average describes nobody’s actual Tuesday. Plans have to survive real Tuesdays.

Not a black box

Every plan can say which limits it satisfied and which trade-offs were taken. If a plan can’t explain itself, it isn’t finished.

Not a promise the world cooperates

Facts change. The commitment is that the plan gets re-checked when they do — not that they won’t.

The name is the method. Axio — from axiom, a foundational truth. Vex — from the Latin vexare: to shake, to agitate, to trouble. Axiovex shakes the axioms: the assumed numbers, the inherited plan, the way it has always been scheduled — until what is actually true falls out. Constraint-based planning is that shake made into a working discipline: no assumption gets built on until it has been tested against the real limits. Then we build on that ground.

Honest limits. The method does not create openings, seats, instructors, or time. It finds the plan that fits what exists, as documented at the data vintage the plan states — and when the plan you want doesn’t exist, it tells you which limits would have to change, which is often the most useful answer available. The software engine behind it (the projected constraint engine, PCE) is under development and being validated; the planning discipline itself is what we practice on every engagement today.

Next step

Bring one decision.

Tell us the decision, who owns it, the data you can authorize and the outcome you need to improve.

Start a conversation