A senior scientist once told us, “If your AI can’t learn from the HTE data in my spreadsheets, I won’t use it.” - Here is what we learned from it

A senior scientist once told us, “If your AI can’t learn from the HTE data in my spreadsheets, I won’t use it.” - Here is what we learned from it:

He was being honest and he was right.

That moment stuck with us. It reflected something we’ve seen time and again while working with over 14 process development teams across pharma and fine chemicals: success with AI isn’t just about having the best algorithm.

It’s about fitting into the reality of how scientists work under pressure - with limited time, partial data, and too many other priorities.

Here are two key lessons we’ve learned along the way:

1. Adoption Isn’t a Technical Problem - It’s a Workflow Problem

Most people assume chemists are skeptical of AI. That’s only partially true. They’re skeptical of extra friction.

When we roll out ReactWise in a new team, the questions we hear aren’t about model architecture or hyperparameters - they’re about how this fits into their day-to-day.

Will it slow them down? Will it require yet another format? Will they need to call IT just to upload data?

That’s why we’ve focused so heavily on usability:

• Upload CSVs or Excel - no formatting headaches
• Get clear, interpretable experiment suggestions - no black-box outputs
• Use visuals that chemists already understand - like surface plots and reaction maps

The lesson?

For adoption to happen, the tool must feel like it belongs on the bench - not like it came from a different universe and that’s what we're doing at ReactWise.

2. ROI Doesn’t Come From Novelty - It Comes From Focus

There’s a lot of talk in the industry about the future: autonomous labs, digital twins, continuous optimization.

These are exciting goals - but they’re not where the most value is generated today.

The biggest returns we’ve seen come from something much simpler: enable scientists to focus on high-value experiments that lead them to their goal faster.

If ReactWise can prevent a team from spending three weeks exploring a dead-end route, or help them reach optimal conditions with 30 experiments instead of 80, that’s an operational win and a business one.

It’s about doing fewer experiments in a smarter manner.

That scientist and his spreadsheet reminded us that tools only matter if they solve the problems people actually face.

We’ve since worked with teams around the world, and the same principles keep proving true:

• Make adoption effortless
• Treat historical data as an asset
• Focus on real-world value, not just technological novelty

If you’re in process development R&D and looking to optimize faster, we’d love to show you how these lessons can play out in your own lab.

‍