Representative, anonymised results from delivered client and internal case studies. Figures reflect specific projects and are not guarantees.
ACE-inhibitor side-chain intermediate
One (S,S) intermediate, three manufacturing scales — discovery, kilo and commercial — each demanding different route economics.
Synthetic strategy
Reductive amination of ethyl 2-oxo-4-phenylbutyrate with L-alanine (NaBH₃CN in the lab; H₂ · Raney Ni at commercial scale). Commercial route reaches dr up to 17:1, 70–85% yield on the cheapest commodity feedstocks.
Lab · kilo · commercial, each scale-matched in a single step.
Endothelin-antagonist pyrimidine core
The automated retrosynthesis returned an 18-step de-novo construction. A classical condensation the engine never proposed collapses it to four.
Synthetic strategy
A Pinner condensation (NaOMe, single step) of diethyl 2-(2-methoxyphenoxy)malonate with a 2-aryl-amidine builds the 4,6-dihydroxy-2-aryl-pyrimidine core directly.
Core built in one Pinner step the engine never surfaced — a curated-chemistry win, not a machine ranking.
Oral protein-degrader (PROTAC)
A costly route with low end-to-end yield. A step-wise cost workbook found the dominant cost driver and a late-stage strategy change tripled the yield.
Synthetic strategy
The linker–warhead intermediate (INT-1) was flagged as the dominant cost driver; a late asymmetric hydrogenation (chiral catalyst · H₂) redesigned the route around a pomalidomide-type CRBN ligand.
A 7-tab cost workbook plus schemes delivered for 3-way route selection.
Ionizable LNP lipid (mRNA delivery)
A branched-tail ionizable lipid for lipid-nanoparticle delivery — convergent fragment coupling, with honest, per-lipid manufacturing-scale caps.
Synthetic strategy
A branched C-tail alcohol and an amino-diol head group joined late by esterification / N-alkylation (convergent, 2 fragments). Clinical-grade lipid kg-feasible; research lipid at gram, research-only scale.
LNP role & formulation panels (not ADME); honest scale caps stated per lipid — breadth beyond classic APIs.
Chiral diamino-ester building block
Two stereocentres and a deceptively hard disconnection. A divergent disconnection map (bond × polarity × chiral source) recovered routes single-polarity thinking misses.
Synthetic strategy
An asymmetric aza-Henry gives the syn β-nitro-α-amino ester from a glycine Schiff methyl ester, then reduction to the diamine. Kinetic resolution plus racemise-and-recycle delivers the free single enantiomer.
1 chiral-pool + 3 resolution / kinetic-resolution routes, correctly chosen for the single enantiomer.