The situation this addresses
A small workshop building bespoke assemblies. Quotes live in one system, jobs on
a magnetic whiteboard, and delivery promises in the sales manager’s head. Each of
those is fine on its own. The gap is that nothing compares committed work against
available capacity, so when two large jobs overlap nobody finds out until the
week it matters.
A business in this position does not need an ERP, and buying one usually makes
things worse. What is missing is narrower than that: seeing what is committed,
what capacity exists, and where the two collide.
How I would build it
The first slice would be read-only, and small enough to be useful in a
fortnight: pull accepted quotes from the system that already holds them, express
each as hours against a work centre, and draw it on a calendar. Nothing is
replaced, nothing is migrated, and if it turns out to be wrong very little has
been spent finding that out.
Read-only first is the important part. It earns trust in the numbers before
anyone is asked to change how they work.
Once it is being used and believed, the obvious next pieces are:
- Editable job stages, so the whiteboard can come down.
- Capacity by work centre, including planned absence, rather than a single
headline figure that hides the actual bottleneck.
- A forward view that flags a clash when a quote is accepted, not when the
material arrives.
- Exports for whichever reports someone is currently rebuilding by hand each
month.
What I would deliberately leave alone
The scheduling decision itself. A system like this should show the collision and
let a person decide what moves. Automating that decision is the fastest way to
lose trust in the tool, and trust is the whole point — a forecast nobody believes
is worse than a whiteboard everybody does.
Status: example use case
This is a worked example, not a delivered project, and there is no client
behind it. It is here because the pattern comes up constantly in small
manufacturing and because it shows how I would sequence the work: the smallest
read-only slice first, and the clever parts only once the boring parts are
trusted.
No figures are quoted for it, because there is nothing to measure.