Back

Flow chemistry or HTE? When machine learning enters the lab, the 'best setup' isn’t one-size-fits-all.

April 30, 2025

Flow chemistry or HTE? When machine learning enters the lab, the 'best setup' isn’t one-size-fits-all. I often get this question from clients and prospects:

"We want to automate our laboratory workflow to leverage new digital tools. Should we set up a flow setup with self-driving capabilities, or go for high-throughput experimentation?"

And my answer is always the same: it depends.

It depends on the scope of your work — and where you are in the drug discovery process.

In discovery chemistry, you're often building large molecule libraries by rapidly exploring different structures.

High-Throughput Experimentation (HTE) is perfect here - you can run up to hundreds of parallel reactions on well plates, varying catalysts, solvents, or additives across a plate to find the best combinations fast.

In process development, the challenge shifts: you're optimizing the synthesis of a specific molecule to prepare for scale-up and manufacturing.

This is where Continuous Flow Chemistry combined with self-driving lab automation can be a game-changer.

Flow chemistry is a technique where chemical reactions are run continuously through small, controlled channels or tubing, allowing precise control over reaction conditions like temperature, pressure, and reaction time.

Here’s a quick breakdown to highlight the two methods in more detail:

Flow Chemistry + Self-Driving Labs

Strengths:

  • Effective integration with Bayesian optimization due to a natural sequential experimental workflow
  • Safe superheating beyond solvent’s boiling point
  • Extremely fast heating and cooling due to the high surface-to-volume ratio, ideal for handling hazardous or sensitive chemistry
  • Easy scale-up via numbering up or maintaining similar reactor mixer geometry
  • Easy integration with process analytical technology (PAT) for real-time, online/inline reaction monitoring

Challenges:

  • Experiments are run one after another, which limits throughput compared to parallel approaches.
  • Switching between different reactants mid-run is possible but operationally more complex.
  • Solid starting materials can be handled but require careful system design to avoid clogging

High-Throughput Experimentation (HTE)

Strengths:

  • Parallel screening of many conditions (24/96/384-well plates) delivers very high throughput
  • Ideal for varying categorical parameters like catalysts, solvents, additives
  • Massive data generation to train ML models for early-stage discovery

Challenges:

  • Reaction time and temperature typically fixed across a plate
  • Careful experimental design needed to avoid sparse or biased data
  • Scaling up successful conditions from micro-scale to production scale requires additional process development work.

In short:

  1. If you want to explore chemical space quickly, HTE might be your method of choice

  2. If you want to optimize and scale a known reaction, flow chemistry shines.

And the most efficient labs use both - first explore broadly with HTE, then translate, fine-tune and scale with flow.

At ReactWise, we provide an AI copilot for chemical process optimization and also help our clients choose the right automation capabilities for their needs.

We integrate seamlessly with equipment for both high-throughput and continuous flow workflows, and we even support batch-to-flow transfer using transfer learning to accelerate your transition.

Ready for the next step in your optimization journey?

Do you have questions, need more information about our chemical process?