July 26, 2021

So, you’ve automated 80% of your scheduling work, and now you want to take the next step towards making it to 100%. Not so fast! First, let’s take a closer look at lab automation and ask: how much is too much?

The Lab of the Future is now

Today, the world is driven by automation. From production lines, to mining, from the YouTube algorithm to resource planning, AI increasingly dominates how we work. But when implementing automation at scale, before asking “how” it’s important to first ask “why”: what is actually the goal of pushing toward greater levels of automation?

By definition, automation will prioritize the replacement of manual processes with actions that can be delegated to machines because the work is either too complex or time-consuming for humans to deliver reliably; often, it is pursued in the name of “efficiency”.

When it comes to resource scheduling (especially in large labs), it is clear that there are tasks for which computers are ideally suited. Sorting through large amounts of data to find the best fit between workload, availability and skills genuinely is more efficient when undertaken by a machine compared to a human (yes, even when equipped with a highly detailed Excel workbook!). But the temptation is to ask, “how can we get rid of all the manual planning and scheduling work and just let the computer do the work for us instead?”

 

Too much of a good thing

Once clients see the benefits of automated scheduling, they often ask an obvious question: “Now that we have automated 80% of the scheduling work, we sure can automate 100% of it as well, right?”

Questions along the lines of:

  • “Can we link our schedule to outlook so it can automatically import meetings & training?”
  • “Can we link it to our HR system so it can automatically synchronize competencies?”
  • “Can we configure the scheduling algorithm in such a way so that it can take into account every possible constraint and can handle every possible exception?”

Although interesting questions (and all technically possible), they should be treated carefully. Pursuing fully automated lab scheduling is a pitfall. It’s a typical case of diminishing returns: once you reach an optimal return on investment, every ‘unit’ of extra automation starts working against you.

So, perhaps the questions should instead be:

  • “Will our automation plans save us hands-on time?”
  • “Should we be striving for optimization rather than automation?”
  • “Can we automate without making everything more complicated?”

 

Does the lab actually save time?

One of the main goals of automating the lab scheduling process is to save hands-on time so that it can be more productively deployed elsewhere. After all, with most of the non-value-added work out of the way, lab supervisors can focus more on what they do best: keeping an eye on the schedule adherence, communicating with the lab staff, identifying further optimizations, etc. Nevertheless, sometimes automation can be pushed too far, resulting in manual work merely being shifted from one task to another task.

In a lab context, asking the question ‘do we save time’ is especially critical when deciding with which systems you want to interface. While integrating with a LIMS, ELN or ERP are no-brainers, other types of integrations require a little more caution. For example, what if we integrate our outlook calendars with the automated schedule algorithm? Sounds like a good idea, right? But how often are we populating our outlook calendar with personal reminders? Placeholders for meetings that might not take place? You could end up with an integration between two systems working against each other. In the end, you are spending more, not less, time maintaining a clean schedule.

Takeaway: try to look beyond the superficial time-savings of certain process automation. Identify potential areas where non-value-added time is just moved elsewhere instead of being reduced.

 

Focus on lab optimization, not on lab automation

Let’s face the facts. Computer algorithms are better than humans when it comes to performing repetitive and standardized tasks. Algorithms output better results in a fraction of the time.

In the context of scheduling, repetitive and standardized tasks take up 80% of time spent. Things like:

  • Extracting all the work that needs to be done in a given week
  • Prioritizing work
  • Grouping samples in campaigns based on due dates, priorities, and constraints
  • Matching test methods with available, competent team members and equipment

These are all perfect candidates for automation. Congratulations, you now have a significantly more efficient schedule, and you’ve reduced total time spent on creating it by 80%!

However, the other 20%, is made up of exceptions, exceptional cases, and unexpected events; this is where the human brain still trumps computer algorithms. To configure algorithms so they can handle every possible exception would constrain so much that you’d simply end up with a schedule equally (or even less) efficient than what your human scheduler would have come up with in the first place.

Takeaway: The sweet spot is to automate repetitive tasks, work according to specific standards of execution and not to demand any strategic or creative thinking. In other words: the machine is there to enhance your work, not to replace you entirely. By taking a “co-bot” approach, you can automate the ‘stupid stuff’ (80%), and your human scheduler handles exceptions and exceptional cases (20%) for optimal results.

 

Is lab automation making our work easier?

Automation has to make your life easier. Be careful whenever you’re deciding what to automate and with which software you want to accomplish that automation. Pushing process automation too far could result in complexity, lots of configuration, and master data you need to maintain for years to come. Probably an obvious comment, but using software created to help you tackle your specific lab challenges is critical to maximizing return on automation without adding complexity.

Takeaway: automation relies heavily on the maintenance of master data, such as custom configuration, special rules, etc. Hence, “saving 30min a week by automating the exceptions, but adding 45min a week for master data maintenance” is a classic case of ‘too much of a good thing’.

 

Conclusion

There is a constant drive towards efficiency in the modern lab, which has resulted in a push for ever-greater levels of automation, howewever, machine automation depends on the rules that you define. The more complex the rules, the less flexible an automated process will be and the more human time will be spent maintaining the rules.

Efficiency is not simply about saving time but is also determined by the effectiveness and productivity of the work being performed. Machines can absolutely perform faster and more efficiently than humans in specific areas – however the priority should always be to automate simple, non-value-added and repetitive tasks because these are exactly where the ingenuity and innovation of the human mind become unstuck.

When it comes to streamlining lab processes – including resource planning and scheduling – the approach should focus on optimisation rather than on automation. This means balancing your machine and human work to achieve the ideal combination of speed and complexity – a co-bot approach. In this way, you free-up more of your skilled workforce to concentrate on the complex and strategic work.

 

Want to see an example of co-bot lab planning in action? Discover BINOCS, the leading resource planning and scheduling solution for life sciences labs.

Author

Adam Lester- George

Consultant

No Comments

Post A Comment