Most automation projects start with a shopping list: a liquid handler, a reader, a robot arm. But a better starting point is the current protocol and an honest look at every step.
The first step in automating a lab workflow is understanding what your process is made of. This step gets skipped more often than any other, because the process feels familiar. Familiarity is not the same as examined.
This guide draws on a Retisoft Inc. webinar on lab software and system selection, led by a guest speaker with more than 25 years of software development and delivery experience. Retisoft’s lab automation services start from the same place: break the process down first, then choose equipment to fit it.
People Also Ask
Process isolation refers to taking a specific workflow, breaking it into individual, definable steps with specific inputs and outputs, and then determining which steps can be automated.
A robot can only do what it is told to do. As such, any decisions that need to be made during process execution must be laid out ahead of time, so it is important to understand and justify every detail of the workflow.
Not necessarily. While some aspects of research are perfect for automation (high-volume processes), others are better done with a mix of automated and manual components (discovery work).
A process that hasn’t been dissected into definable steps cannot be automated, because there are no instructions for the workstation to follow. Even if a robot performs a task that seems to require judgment (like rejecting a plate or following up on a particular reading), a human must have written down the instructions for what the robot should do in that scenario. As the speaker in the webinar noted, a robot essentially only does what it is told to do.
A habit not formalized into instructions is an instruction the robot will not follow. On a human bench, the scientist might do whatever is necessary to get a result, without thinking about the implications. On an automated bench, the process will simply stop and wait for instructions.
The first rule of adding automation or really any new software is to accept that your process will change. Hand-built flows often include steps that no longer serve a purpose. They were either required by a previous system that is no longer in use or were steps leading to a different end goal than the one you want now.
Maybe the lab still does a bunch of stuff because an old software required it, but that software is no longer in use. Or maybe some steps used to be a prerequisite for another end goal and nobody removed them when that end goal changed.
But the most common reason I’ve seen is that somebody did it one way and taught everybody else to do it that way.
If you’re doing a process review, ask yourself why we do each step. The most common answer I’ve gotten is something like, “Well, we’ve always done it that way.” That’s a red flag. Everything in an automated process should have a clear justification.
The webinar called this process isolation: breaking larger work into its smallest parts. A plate read is a good example of a single task. What matters is the result, not whether a scientist or a robot carried the plate. Two details still matter, though: when the read happened and how. The method becomes part of the data, not just the final value.
A number of lab operations consist of a short run of two, three, or five tasks that are repeated throughout a workflow, for example, adding a reagent, washing, and reading.
Define a group once as a subprocess, and you can reuse it again and again. The individual tasks within a group do not always have to be in a particular order.
Every workflow is a long string of small steps, each of which is in a specific order, so a 4-step group can create multiple variants, and changing a single task, such as a washing method, means searching through every variant that uses it. The speaker called this complexity snowballing.
Once subprocesses exist, a large workflow becomes a linear series of them. This is the foundation of integrated laboratory automation: fewer permutations, clearer handoffs between people and instruments, and far less maintenance.
You validate a task once, and you know it works in every subprocess and workflow that uses it. Without these structures, every change to a task requires retesting every large process in which it occurs.
Automation does the task as designed, so subtleties in the design can throw off the whole process. The speaker tells the story of a lab that switched to new barcoded tubes from a different vendor that had to be ordered in bulk. The new tubes had a 30 percent failure rate for reading the barcode. He explains it’s all about the physics of how the light refracts and how the camera and software at the other end read the barcode. You have to test every change, even something as seemingly simple as a change in consumables. Consistent inputs let lab robot software reduce error run after run.
Simplicity isn’t an end in itself, but flexibility stops helping at some point. Giving users a half-dozen options at each turn leads to fiddling, changing the data stream, and making everything inconsistent. To keep runs comparable, keep it simple.
A well-broken-down workflow maps almost directly onto good laboratory workflow automation solutions. Genera, Retisoft’s scheduling software, shows how. Not every platform handles that mapping the same way, as our comparison of automation-focused and equipment-centric platforms shows.
In Genera, each lane represents one labware item, and each subprocess is a reusable unit within the overall process. In Retisoft’s ELISA webinar example, the main process contained three subprocesses: adding samples and standards, plate washing, and stopping the reaction before the read.
The complete assay version began with a decision node that asked whether the plate was precoated and skipped the coating steps in the latter case. Wash and incubation steps are nested within each stage as subprocesses. Lanes, decision nodes, and loops are explained in full in how Genera builds automated processes.
Recycling previous work is where Genera shines. Parameter variables let you reuse a single process in different scenarios, so a scientist who needs 5 wash cycles and another who needs 10 can share the same process. Reuse like this is central to how Genera takes over repetitive scheduling tasks.
Handlers and variables are copied from one process to the next. You can also import entire workspaces, including labware definitions and zones, from another Genera instance.
Try this with one workflow before talking to any vendor:
Where full automation does fit, the benefits of a fully automated laboratory build quickly.
Automating a lab workflow well starts long before any equipment arrives. Question every step, isolate single tasks, group them into subprocesses, test the small details and keep each task simple. Equipment choices become clearer, and the automated process stays easier to maintain.
Retisoft Inc. follows the same approach in its lab automation services, starting with process isolation and fit-for-purpose equipment selection, built around the instruments you already own. Book a Genera demo and bring your workflow breakdown with you.
Break the workflow down. List every step, confirm why each exists, and isolate single tasks before selecting instruments or software.
A short, reusable group of tasks, such as adding a reagent and washing, defined once and used across many larger workflows.
Automation performs tasks exactly as designed. A new tube or plate type changes conditions such as barcode readability, so every small change needs testing.
Not necessarily. High-volume work often suits full automation, while discovery work benefits from a mix of manual and automated steps.
Genera models workflows as lanes and reusable subprocesses, with parameter variables, decision nodes, and importable workspaces, so teams avoid rebuilding the same logic.
Latest Blogs