One of our big motivations for building Sphinx was seeing lab after lab where data prep had become a hidden bottleneck to scientific progress. After weeks in the lab running an experiment to answer their most pressing questions, scientists in these labs would wait to look at the data until they had a full day to devote to analysis. Sometimes that might be the next day. Other times it was longer.

But even those full days weren’t spent on deep analysis. Most of the time went to copying and pasting between spreadsheets to get the data into a form where analysis could even start. No wonder these scientists were putting it off.

And while a day or two of delays may not seem like a lot, they add up. When planning the next experiment depends on understanding the last one, an extra day for each experiment translates to weeks and months.

But the even bigger issue was the missed opportunity to dig deeper. By the time scientists had gotten to a point where they could do the basic analysis, they didn’t have the time or energy to dig much deeper. After the huge investment in generating data in the lab, the cost of data prep was stopping them from getting their money’s worth.

The most frustrating thing about all this was that the data prep wasn’t even very complicated. It took a long time because it was tedious. But it tended to be the same thing over and over again. It seemed like an obvious candidate for automation.

However, there were a few common reasons why it wasn’t being automated: 

• The scientists used spreadsheets that they were comfortable with, but which weren’t designed for automation. 
• The computational teams weren’t big enough to automate all the different assays. 
• And even if they were, there were already minor changes from experiment to experiment that were just big enough to break any attempt at automation.

We built Sphinx to slash the amount of time scientists spend on data prep by addressing these issues.

First we made it easy to create data prep and analysis templates and share them with scientists. This allows computational biologists and data scientists to focus on the analysis instead of the machinery to automate it. In startups that have adopted Sphinx, computational teams are the heroes that save the bench scientists from tedious data prep. And they still have time left to work on more complex problems.

But it wasn’t enough just to help computational teams automate faster. We still faced the problem of managing the variance from experiment to experiment. Computational teams could add more parameters to account for this, but it means more complexity for users. If you add too many parameters, bench scientists will just go back to using Excel.

We addressed this issue by introducing AI to help scientists manage the added complexity that inevitably comes with increased flexibility. Sphinx’s AI agent, Metis, reviews data to identify its structure and transform it. Then it helps bench scientists configure analysis templates and visualizations. This significantly reduces the learning curve without sacrificing scientific rigor.

Between the templates and our AI agent, Sphinx turns what used to be hours of manual analysis into minutes. And we’ve seen it fundamentally change how our customers work: Scientists who use Sphinx do their initial analysis as soon as the readout is complete. Then on those days they set aside for analysis, they dig deeper into the data or head back into the lab.

Because these scientists can spend their time on deeper insights, they make better decisions with more confidence. Without the hidden bottleneck of data prep, bench teams that use Sphinx can finally work on what matters. 

‍