Find the loss, run the project, measure the gain. All from the same machine data.
MACH CI Projects tracks continuous improvement projects against a baseline locked from machine data, shows whether each change is meaningful and what it’s worth, and keeps checking that the gain holds. You can see the line running better. Then someone asks how much better, compared to what, and whether it'll last, and the answer is a spreadsheet only its author trusts.
Last updated: October 13, 2026
Virgil finds the losses that keep coming back. Then offers to make each one a project.
Ask Virgil a question on any dashboard, and when the answer turns up a process loss worth fixing, like changeovers running long across a line, it offers to start a CI project for it, with the evidence alongside. One click opens the setup already filled in from the conversation, and if a project already covers that loss, Virgil points you to it.
Changeovers on Presses 1 to 3 average 42 minutes, 14 minutes slower than the best shift on the same parts. That’s about 9 hours of press time a week.
Set it up- Virgil can also review your recent data and suggest the projects most worth running, each with the numbers behind it.
- Once a project is running, ask Virgil how it's going or what to try next.
Set up a CI project from the problem. The baseline numbers are already there.
Pick what you're going after, choose the machines, and say when the change goes in. Each goal shows you its real starting value from the machine data before you set a target, along with roughly how many days of results it'll take to tell. You can set a target without running a time study first.
Shorten changeovers.
Measured from the changeover timer already running at every machine.
Reduce breakdowns.
Track how much time breakdowns cost, how often they happen, and how fast they’re fixed.
Fix the problem that keeps coming back.
Go after a downtime reason that shows up across a line or plant.
Cut scrap and defects.
Measured from the rejects operators log against each order.
Run it with the method your team already uses, like SMED, A3, PDCA, DMAIC, or a kaizen event, and the project board follows its phases. Templates also cover line balancing, 5S, and safety, and your team can write its own.
- Goals can be any KPI you track, including downtime, schedule attainment, a formula of your own, or a score your team records by hand, like a 5S audit.
- Mark a goal "must not get worse," and MACH watches it alongside the main goal, so if a faster changeover starts costing you scrap, you’ll see it on the project.
An action board for every CI project. Every step has an owner and a due date.
Every project has an action board with owners, due dates, and the phases of the method you chose, so the plan lives with the project, not in a separate spreadsheet.
Overdue steps can send a reminder to their owner.
A baseline locked from machine data. Every result measured against it.
When you start tracking, MACH locks the baseline from the period before the change. Every period after is measured against it from the same machine data your dashboards use, showing how much things changed and whether the change is meaningful or just the normal ups and downs. Until it's clear, the verdict reads "can't tell yet."
| Goal | Baseline | Sep | Verdict |
|---|---|---|---|
| Average changeover | 42 min | 31 min−26% · meaningful | Target met |
| Changeovers per day | 5.1 | 5.4no target, tracked | Tracked |
| Scrap rate, must not get worse | 3.6% | 3.5%guardrail | Held |
11 minutes saved per changeover × 5.4 a day × your $135 machine-hour rate. Confirmed.
Presses 4 and 5 improved 1.2 minutes over the same months. The result above is net of that.
Here's what you're seeing: an illustrative changeover project two months in, with the gain confirmed as meaningful and the scrap guardrail holding.
- Each goal gets a plain verdict: target met, improved, can’t tell yet, worse, or not enough data yet.
- Name comparison machines the change didn't touch, and the result is reported net of whatever happened to them, so a good month for the whole plant isn't credited to your project.
Connects to: the KPIs you've defined and the downtime reasons operators log at the machine.
Turn the improvement into dollars. Using your own cost numbers.
Enter your own assumptions once, like downtime cost per hour or scrap cost per unit. Each period then shows the savings and an annualized figure, with the formula and inputs right underneath. Only meaningful changes count as confirmed savings, so the total you present is one you can defend.
One page to show leadership.
Every project gets its own dashboard with the before-and-after, the savings, and whether the gain is holding.
Every project at once.
See confirmed savings across all your projects, which gains are holding, and which projects are falling behind.
Pick up where this one stopped.
Spin off a follow-up project and it carries the goals, machines, and cost assumptions forward.
Sustain tracking after the project closes. MACH is still watching.
Most gains fade. The project closes, the team moves on, and a few months later the changeover is creeping back toward where it started. When you complete a project in MACH, it keeps comparing current performance against the baseline and the result you achieved, on a schedule you set.
Holding.
Still at or near what the project achieved, so there’s nothing to do.
Slipping.
Part of the gain has slipped away, past the threshold you set. Virgil adds a short note on what changed.
Regressed.
Back where it started. You find out from MACH instead of from the next bad month.
- A change in status shows up in Hub, and an optional weekly email rolls up every closed project in one place.
Questions about CI Projects.
How does MACH measure whether a CI project worked?
When you start tracking, MACH locks a baseline from the machines' own data before the change. Every period after is measured against it from the same data, showing how much things changed and whether the change is meaningful, with a plain verdict: target met, improved, can't tell yet, worse, or not enough data yet.
What if the whole plant improved for some other reason?
Name comparison machines the change didn't touch. MACH reports the result net of whatever happened on those machines over the same periods, so a good month for the whole plant isn't credited to your project.
Does MACH put a dollar value on an improvement?
Yes. You enter your own cost assumptions, such as downtime cost per hour or scrap cost per unit, and each period shows the savings and an annualized figure, with the formula and inputs. Only meaningful changes count as confirmed savings.
What happens after a CI project is closed?
MACH keeps checking. On a schedule you set, it compares current performance against the baseline and the result you achieved and marks the project holding, slipping, or regressed. A change in status shows up in Hub, and an optional weekly email rolls up every project's status.
Which improvement methods does it support?
Run a project with the method your team already uses, like SMED, A3, PDCA, DMAIC, 5S, or a kaizen event, and the action board follows its phases. The measurement works the same way for all of them.
What does Virgil do in a CI project?
It suggests projects from your recent data, offers to start one when an answer turns up a loss worth fixing, and tells you how a running project is going. Your team saves every project and every step.
Is CI Projects priced separately?
No. CI Projects is part of Reporting & Analytics, which comes with every MACH product: Monitor, Schedule, or Complete. See pricing.
Your next improvement project needs a baseline. Your machines already have one.
We start with a Shop Floor Diagnostic. One-hour call to understand your situation, then a written assessment of where the losses are and what closing them would be worth. Where we can't help, we say so. CI Projects comes with every MACH product.
CI Projects is part of Reporting & Analytics, included with every MACH product at no additional charge. See pricing.