The Dark Side of Contract Development

I spoke remotely with a software company.

They are developing software such as image processing on consignment.
The development environment and equipment are basically specified by the client.
We felt that they have a wide range of knowledge and experience to deal with these issues.

On the other hand...I think the work must be hard.

I can imagine the path I have been on, too.

It may be limited to contract software development for automation of visual inspection.

Since the goal is to "replace people," the budget cannot exceed the labor cost of inspectors, which can be reduced.
The objects to be inspected are varied and one-size-fits-all; if 100 projects are commissioned, 100 different software packages will need to be maintained.
Because of the automation of human senses, quantitative specifications often do not exist. Even if there were, the specifications often do not work.
Even if excellent software is created, it is difficult to deploy it due to confidentiality restrictions.

This is not sustainable...the company and the engineers will collapse.

In our case,

Early on, a general-purpose visual inspection method was compiled into the "standard software FlexInspector".
Customer requirements were met by extending the standard software. The software became more and more versatile.
This allowed us to spare manpower for software development.
The ability to check the detection capability in advance has reduced the amount of follow-up work after the introduction of the software.
The repeat rate has increased to more than 80% as a result of the proven results at the customer's site.
By selling the hardware as a set, support became easier, and profits increased.
By providing in-house support for equipment that has a large return on investment, the company was able to further increase profits.
Because of the high return on investment of the equipment, repeat orders increased.
The profits earned were enough to cover the cost of the next technological development.

and over a period of 20 years.

I think again that the biggest thing was to move away from contracted development.