Writing.
Essays on AI, data integrity and the people side of change, from someone who has done the work. New pieces appear here first.
16 October 2026 · AI drug discovery · 2 min read
Protein design moved to the atom. The bottleneck moved to the bench.
A new system from the University of Washington designs enzymes atom by atom, which changes the question an investor should ask a drug-discovery founder.
21 October 2026 · Operations · 2 min read
The quiet half of biotech: scaling it next to a sugar mill
A Brevel and Wave Group plant jumps from a 5,000-litre tank to 300,000 litres, on a feedstock sitting next door. Scale-up is the half that decides whether biology reaches a shelf.
20 October 2026 · AI drug discovery · 2 min read
A model that reads the genome one cell at a time
Arc Institute's Bolero turns years of cell atlases from a catalog into something you can question.
20 October 2026 · Healthtech · 2 min read
The $60 genome, and why the cheap instrument sets the ceiling on the models
Ultima Genomics says it can sequence a whole human genome for $60. The interesting consequence is upstream of medicine.
20 October 2026 · Operations · 2 min read
When the measuring gets cheap, the thinking becomes the constraint
A Tohoku University system runs a full metals experiment without a person, more than ten times faster than by hand. The scarce resource becomes deciding what to test next.
20 October 2026 · AI drug discovery · 2 min read
An AI ran a whole protein-design campaign on its own
In 64 hours, a model drove a dozen design tools, screened half a million candidates and handed back 17 binders. The design step is becoming something you point at a problem and leave running.
18 October 2026 · DeepTech · 2 min read
Searching for materials in plain English
An MIT model pairs the 3D structure of a crystal with a written description, so you can query the materials space with words.
14 October 2026 · Trust and provenance · 2 min read
Google opened an error-checker for technical documents. Peer review is just the first customer.
Deep Review stress-tests a document for logical gaps and math errors. The same problem shows up in every due-diligence memo and regulatory filing.
14 October 2026 · Trust and provenance · 2 min read
An AI detector that reads the ideas, not the prose
A University of Maryland and Google DeepMind detector asks whose thinking a document contains, which reaches well past the classroom.
12 October 2026 · AI drug discovery · 2 min read
Mapping the layer of biology where a lot of disease actually lives
An Arc Institute atlas maps where 67 proteins grab onto RNA, opening the least-charted of biology's three control layers.
July 2026 · Healthtech · 8 min read
eCOA vs ePRO: a practitioner's guide
What electronic clinical outcome assessment (eCOA) and electronic patient-reported outcome (ePRO) really mean, how they differ, and what to ask before you pick a platform.
June 2026 · AI drug discovery · 9 min read
AI in drug development: a pharmacist's view on data integrity
Where AI moves discovery forward, where data drift poisons it, and what a pharmacist's eye catches.
Short notes on LinkedIn
- September 2026 · AI drug discoveryWhat a pharmacist sees in the AI pipelineThe habits that keep data honest when the model moves faster than the audit trail.
- September 2026 · LeadershipOn the next wave of AI in the enterpriseA short reflection on what changes when the tool becomes part of the team.
- September 2026 · LeadershipTrust, speed, and the hard part of scaleWhy the organisations that win slow down in the right places.
- March 2026 · OperationsHiring the right softwareChoosing tools the way you'd choose a colleague: for judgement under pressure.
- March 2026 · CommunicationsGenerative AI at workWhere it changes the day and where it just adds a tab. Notes from inside large enterprises.
Field notes
- 01
Intelligence is the commodity now. The system is the edge.
- 02
Anyone can buy the engine. Few rebuild the aircraft around it.
- 03
The work is in the almost.
- 04
The number on the slide isn't the number in the world.
- 05
Most reports don't break. They quietly stop being true.
- 06
Models don't fail in production. Trust does.
- 07
Accuracy is cheap. A decision that changes is expensive.
- 08
The job is to make the moment of commitment quiet.
Field notes by email.
A new essay about once a month on AI, data integrity and the people side of change.