Chemdem uses AI to predict reaction success and yield before you even start — powered by real squarate-amide data and IBM RXN.
No chemistry PhD required. Enter your amine, get a prediction in seconds.
Type any amine by name (e.g. "4-fluoroaniline"), CAS number, or SMILES string. Chemdem resolves it via PubChem and OPSIN automatically — no manual structure drawing needed.
Chemdem uses Morgan fingerprint similarity matching and a k-NN model trained on 42 real reactions from the Wren Group (SynOpen 2023). It checks for hard fails, steric effects, nucleophilicity, and catalyst requirements.
Receive a predicted yield (%), confidence score, 2D product structure, likely side products with risk factors, and references to the closest real reactions in the dataset — all in under 2 seconds.
Designed for researchers, students, and lab practitioners.
Trained on 42 experimentally confirmed squaric acid monoamide reactions from published Wren Group papers. Includes 2 confirmed hard-fail reactions for reliable negative predictions.
Predicts likely side products for every reaction: bis-squaramide double substitution, squaric acid hydrolysis, unreacted DES, and N-oxide formation — each with risk level and mitigation advice.
Every prediction comes with a confidence level (High / Medium / Low) based on Tanimoto similarity to known reactions. Exact dataset matches are clearly labelled as reported yields, not estimates.
Enter any compound by IUPAC name, common name, CAS number, or SMILES. Chemdem resolves structures via OPSIN and PubChem, cross-checks them, and flags any discrepancies automatically.
Instantly identifies experimentally confirmed no-reaction conditions: ortho-chloroaniline, ortho-bromoaniline, and unprotected piperazine — even at reflux with catalyst. No wasted lab time.
Every prediction links to the closest real reactions in the dataset, sourced from Long et al. SynOpen 2023 and Bioorg. Med. Chem. 2024 — so you always know the evidence behind the prediction.
Chemdem is grounded in real experimental data from the Wren Group — not synthetic or scraped data.
Aniline and benzylamine series. 29 reactions including 2 confirmed hard fails (ortho-Cl and ortho-Br anilines). Establishes the key SAR: para > meta > ortho for yield.
Heterocyclic amine series. 13 reactions including morpholine, piperidine, and cyclic amines. Includes HDAC8 biological activity data — best compound: para-F aniline (IC₅₀ = 1.20 μM).
Zn(OTf)₂ catalyst is required for anilines but not benzylamines. Reflux can boost yield dramatically (up to 99%). Ortho-substituted anilines consistently fail due to steric crowding.
Patterns extracted from 42 real squaric acid monoamide reactions.
ortho-Cl and ortho-Br anilines → NO REACTION even at reflux + 48h
Stop guessing. Stop wasting reagents on reactions that won't work. Let Chemdem screen your squaric acid chemistry before you set foot in the lab.
University research project · Free to use · No account required