Log in Intro call

Insights

Insights

Articles for analytical, QC and CMC teams: spectra, impurities, stability, ICH M7 and HPLC, each with a worked example.

Model Cards Under ICH Q2(R2) and Q14

Guide8 min read

Model Cards Under ICH Q2(R2) and Q14

A model card that satisfies a GxP reviewer: intended use, training and test data, performance with its split, domain limits and versions.

October 27, 2026

Benchmark7 min read
Solubility prediction benchmarked: what the numbers say, and what they do not
On the public FastSolv benchmark our solubility model reaches an RMSE of 0.70 (SolProp) and 0.87 (Leeds) log units for compounds not in the training data. Every prediction carries a 90 % interval. What that means for solvent screening, and where a measurement still has to decide.
October 27, 2026
Guide9 min read
Antibody Developability Across Eight Risk Dimensions
The eight developability risks to screen before a lead decision: aggregation, viscosity, solubility, charge, chemical liabilities, immunogenicity.
October 23, 2026
Guide9 min read
An Audit Trail for AI Results in the GxP Lab
What a reviewer needs to accept a computed result in a GxP lab: the record fields, versioning of model and data, and review by exception.
October 20, 2026
Guide8 min read
Protein Structure Prediction for Small Biotech Teams
What a predicted protein structure is good for without a structural biology group: the questions it answers, the ones it does not, and how to check a model.
October 16, 2026
Explainer8 min read
Ecotox Pre-Assessment From Structure: What In Silico Can and Cannot Do
Where a non-testing estimate of environmental endpoints is accepted, what documentation it needs, and which decisions still require a study.
October 14, 2026
Explainer8 min read
How Much Data Does a Custom Chemistry Model Need?
The honest answer is a learning curve, not a number: how to measure what your data buy, and which dataset sizes are realistic.
October 12, 2026
Guide8 min read
Docking Scores Are Rankings, Not Affinities
How to use docking scores for what they are good at: setting up a screen whose ranking you can check, picking a cut, and reporting the result honestly.
October 7, 2026
Guide8 min read
Closing Data Gaps in Safety Data Sheets
Which safety data sheet gaps can be closed with a calculated value, how to label one so it is not read as measured, and which entries need a study.
October 5, 2026
Explainer8 min read
Peptide Identity From MS/MS: b and y Ions Explained
How b and y ions confirm a peptide sequence from MS/MS: mass rules, a worked tryptic example and the gaps that leave identity open.
October 1, 2026
Explainer8 min read
When a Model on Your Own Data Beats Public Models
Four conditions under which a model trained on in-house data beats a public one, three under which it does not, and the comparison between them.
September 28, 2026
Guide7 min read
Where DMTA Cycles Lose Time
DMTA cycles rarely lose time in synthesis. They lose it in queues, in data that arrives unusable, and in decisions nobody is scheduled to make.
September 24, 2026
Guide7 min read
E-Factor and PMI: Calculating Them From Your Mass Balance
E-factor and PMI from the data you already have: the formulas, a worked single step, how solvent recovery changes the number and what the metrics do not cover.
September 22, 2026

Your question, evaluated for your lab.

30 minutes on Microsoft Teams with the CovaSyn team. If it is not a fit, we will say so.

Intro call See an example evaluation