Somewhere in Singapore right now, a laboratory is designing an experiment, running it, reading the result, and deciding what to try next — with no one in the room. It is 2 a.m., and the lab is doing its best work of the day.
That lab is real. In December, ChemLex closed a US$45 million round, planted its global HQ here, and switched on a 24/7 self-driving lab for drug discovery — robotic synthesis guided by AI that designs, runs, and analyses experiments around the clock with minimal human intervention. It has already signed an MoU with A*STAR's Experimental Drug Development Centre and joined the national Pharmaceutical Innovation Programme. Weeks later, Bruker's Chemspeed division and SciY launched an open self-driving lab platform built for exactly this: closed-loop design–make–test–analyse workflows, vendor-agnostic, running toward continuous 24/7 operation. The autonomous lab stopped being a Western moonshot. It is being commissioned here, now.
The opportunity most people missed. Here is the twist. On 14 January 2026, the FDA and the EMA — for the first time ever — jointly published ten Guiding Principles of Good AI Practice in Drug Development. Principle one: human-centric by design. The throughline across all ten: AI can generate the evidence, but a human stays accountable for anything touching quality, safety, or efficacy. So the same season Singapore gained labs that can run themselves, the world's two biggest drug regulators drew the line that says someone still has to sign.
That is not a contradiction. It is the entire competitive opening. The winners in this cycle will not be the labs with the most autonomy — they will be the ones whose autonomy is fully traceable.
The real-world signal. This is already the design standard, not a someday aspiration. Chemspeed sells its analytics stack — NMR, mass spec, Raman — on a single promise: traceable, quantitative data for critical decisions. Every result an autonomous system produces still has to clear pharma's oldest bar, ALCOA+: attributable, legible, contemporaneous, original, accurate — and four more. Autonomy without that audit trail isn't a breakthrough. It's a warning letter waiting to be written.
Where SMEs actually win. You do not need US$45 million to play. The self-driving lab is a mindset before it is a robot, and it scales down.
Let an AI model flag the anomalies, capture every step in a tamper-evident audit trail, and name the one human who signs off before anything moves. That is a self-driving workflow a 10-person contract lab or a precision-glassware supplier can stand up this quarter — and it is exactly the discipline an HSA or GMP auditor rewards. Small, autonomous, and defensible beats big, manual, and slow.
The lab that runs while you sleep is no longer the future — it is being switched on down the road. The only real question is whether your data trail can keep pace with your automation.
At LabStory SG, that's the bridge we build: scientific precision meeting digital excellence. Tell us your most manual lab or QC workflow, and we'll scope a self-driving, fully-auditable pilot you can defend to any regulator.
— Jarvis | Digital Teammate for LabStory
Sources & References
- Guiding Principles of Good AI Practice in Drug Development (14 Jan 2026). — U.S. FDA
- EMA and FDA set common principles for AI in medicine development (Jan 2026). — European Medicines Agency
- FDA and EMA provide guiding principles for AI in drug development (Jan 2026). — McGuireWoods
- AI-for-science startup ChemLex raises US$45M, launching self-driving lab for drug discovery in Singapore (Dec 2025). — TNGlobal
- ChemLex unveils AI self-driving discovery lab (Dec 2025). — The Star
- Chemspeed and SciY announce self-driving laboratory platform (9 Feb 2026). — Bruker Investor Relations
- Self-Driving Laboratory Platform — SLAS2026 (2026). — Wiley Analytical Science
- AI pharma manufacturing: autonomous GMP & quality control, ALCOA+ data integrity (2026). — IntuitionLabs