What I Learned After Two Weeks of Conversations About Chemistry Automation

A couple of weeks ago I wrote about hearing the same line over and over:

"Chemistry automation is still five to ten years behind where it needs to be."

Since then, I've been on a bit of a mission.

I've talked to people who run automated chemistry labs. People like Matt Smith at Labman and Bernd Martensen at Zinsser Analytic, whose teams not only build custom instruments for chemistry automation, but also ship their own control software on top.

I've also had a bunch of DMs asking me to share what I've learned "from the basics up."

So here's my current view and why it finally makes sense to me that chemistry automation is lagging biology.

1. Compared to other industries, lab automation is still the wild west

When I first came into lab automation, I was shocked by how it compares to car plants or factory automation:

• No real standards
• Fragmented hardware and software
• No clear "platform winner"

That's not my opinion, that's a consistent message from people who've been building this industry for decades.

Biology automation isn't perfect, but it's much further along than chemistry.

2. Chemistry is messy in ways that matter for automation

Once you step into chemistry automation, the complexity ramps quickly:

• You're not just moving microplates. You've got vials and flasks of every shape and size, in glass or plastic, sometimes holding microliters and sometimes a liter or more.
• You're handling powders and solids, not just liquids. Getting the right mass of a solid into a vessel is trickier than people think, a few milligrams over or under can matter a lot.
• You're dealing with solvents, corrosive reagents and gases. You often need inert gases (like nitrogen or argon) so your reaction doesn't quietly die because of oxygen or moisture.

Funny enough, things don't actually "blow up" very often. Tiny explosions are rare. But chemistry does have more edge cases and more ways for things to quietly go wrong.

Even simple decisions aren't simple:

• Do you use plastic tips, stainless steel needles, or ceramic/glass tips so your solvent or reagent doesn't attack your hardware?
• Do you integrate balances so solids are weighed automatically, or do you keep weighing as a separate manual step? For some labs, weighing is "hidden" inside an instrument; for others it's front-and-center and still very manual.

All of this means there's still a lot of trial and error in chemistry automation, and different teams are re-solving the same problems over and over.

And there's a human cost: the "automation hacker" who builds and maintains everything becomes irreplaceable. Their career progression slows because they're now the person who keeps the system running.

3. Why biology automation people have it (relatively) easier

In biology, automation is further along, so people know a few things:

• There's usually an off-the-shelf instrument or some combination of plate-based devices that will do what they need.
• There are multiple lab automation software options (none perfect, which is why we built Lab Donkey), but at least there's a market.
• The skill set of "automation scientist" is more recognized, and doesn't automatically trap you in a lifetime of babysitting a single custom system. For those trapped, call me.

In chemistry, that's not true yet.

If you want serious automation, you either:

1. Build your own robotic setup from scratch, or
2. Go to companies like Labman, Zinsser or Chemspeed and have them build a custom system.

Those teams do great work, but custom hardware comes with higher cost, longer lead time, and more risk that only your internal wizard knows how it truly works.

4. The three ingredients for successful chemistry automation

What became clear over the past couple of weeks is that chemistry and biology share the same three pillars:

The science itself

The chemistry is the hard part intellectually, but the scientists largely have that covered. The fear comes from automating chemistry. When people imagine things going "POOF", some automation vendors simply refuse to get involved.

The instruments

In biology, most of what you need already exists as catalog products, and you can often mix and match instruments to build a workflow. In chemistry, the "standard" instruments often don't exist. That's why companies like Labman, Zinsser, Chemspeed and others end up building custom platforms that fit very specific chemistries and workflows.

The orchestration software

In biology, you have several general-purpose scheduling/orchestration tools. They're mostly built around plates, wells and liquids. In chemistry, those same tools often fall over. They were never designed for workflows built around vials and flasks that need to be split, merged, transported, weighed, heated, cooled, gassed, scanned, with both solids and liquids in play.

When I spoke to people using the leading biology automation software in chemistry labs, a pattern emerged:

• The software has been retrofitted for chemistry.
• The original data model and assumptions are plate-centric.
• That makes it almost impossible to get the flexibility that a chemistry lab actually needs.

That's exactly the gap I wanted to test against Lab Donkey.

5. Where Lab Donkey fits

I walked Lab Donkey through our model with people who live and breathe chemistry automation.

The conclusion we came to together:

• We can already handle a surprising amount of chemistry automation, and we're starting closer to the target than most biology-first tools.
• Chemistry is hard, but chemists have the science. Instruments are hard, but companies like Labman, Zinsser, Chemspeed and others have shown they can build what's needed.

What's been missing is standard, drag-and-drop, customizable, extensible software that can:

• Treat vials, flasks, reactors and plates as first-class citizens
• Orchestrate solids, liquids, gases and environmental controls
• Integrate with both off-the-shelf and custom hardware

That's the problem Lab Donkey is built to solve.

So we're going to jump in with both feet and start bidding on chemistry projects, alongside the innovators who are already building the hardware side.

6. One last point: nobody should try to do it all alone

The biggest failure mode I see in chemistry automation is when one company tries to own everything:

• The science
• The custom instruments
• The integration
• The software stack

That usually leads to brittle, bespoke systems that nobody else can support.

Partnership is the answer: Chemists doing world-class science. Specialist hardware companies building world-class instruments. And a neutral software layer that can orchestrate all of it.

That's where I want Lab Donkey to sit.