There is a dramatic difference between full automation and partial automation. When I was automating protocols with the OT2 robots, I quickly realized that I could only manage a few robots at a time (in my case, 3 robots). There was actually a lot of work adjacent to my protocols that I did not realize really existed: labeling plates took a long time, sealing plates took even longer, and loading/unloading pipette tips took more time than I thought it would. There exists a ratio for every automation workflow of “people per unit productivity”, and this ratio dictates how effective the automation system is.
Most importantly, I do not believe this function scales linearly - as you add more people to your organization, the cost per person goes up in non-trivial ways. Perhaps not in salaries, but in culture around your company - you lose the ability to be small and agile. While I’m sure there are business-side solutions to this problem, I’ve observed that large companies are not only less efficient, but less free.
The real question is - how do you decrease people per unit productivity given our current automation landscape?
We currently have many robots in the market to do specific tasks. Usually, these robots are loaded by lab technicians, which then do their specified tasks, and then the lab technicians move the plates/tubes to the next machine to do the next specific task. This movement, the transfer process of bringing materials from one place to another, is primary problem preventing us from decreasing the amount of people needed in this process. From the low-end of the market, where I exist, I find that there are solutions to all problems/tasks except for the physical movement of materials.
This is the problem I call “the transfer problem”. So long as you have all the basic machines necessary to complete a workflow, a solution to the transfer problem should allow you to set reagents in one location once a day while allowing your entire automation workflow to procede along continuously, 24/7, without people.
There are many ways to solve the transfer problem (and I would highly recommend playing the game “factorio” to learn more about solving them). I am using an AR3 robotic arm for local transfer and roombas to transfer materials between remote places. Given enough software, this hardware should be sufficient to create a lab that has solved the transfer problem.
However, I would like to emphasize reality: you do not have to solve this problem to be a successful business or successful lab. In fact, it may be worthwhile to invest your time into different endeavors. But the time for automated labs is coming. Keep that in mind. They will be more efficient.
Keoni Gandall