A combinatorial space your patent counsel can't enumerate, scored against the prior art, optimised against vulnerability.
A human cabinet analyses two to five readings of a claim. The Lattice enumerates the entire space — six axes, every jurisdiction, every jurisprudence era — and emits the wording with the lowest cross-juridictional vulnerability.
Module spec'd · MVP not yet implemented · Q3 2026
Every feature of every claim is read along all six axes simultaneously. The cartesian product is the lattice.
Open-ended vs partial-closed transitions, range endpoints, 'about X' tolerance windows, wherein-clause semantics.
Festo / Warner-Jenkinson insubstantial differences — bioisosteres, salts, polymorphs, isomers, isotopes, prodrugs.
RDKit + bioisostere KG sub-genus enumeration. Patani & LaVoie, Meanwell — chemical graph traversal.
Broad genus → preferred → picture-claim cascade with DOE-extension envelopes per jurisdiction.
Phillips v. AWH (Fed. Cir.), G 1/24 (EPO/UPC), purposive construction (UK), JPO restrictive reading, CNIPA variable.
Pre/post G 2/21 plausibility, G 1/23 non-reproducible art, In re Cellect ODP×PTA, USPTO Inventorship Nov 2025.
Calibrated against PTAB IPR + EPO Boards of Appeal datasets. Output: calibrated probabilities, not gut feeling.
Does the prior art literally disclose every element of the interpretation? XGBoost + LegalBERT, calibrated on PTAB IPR + EPO opposition decisions.
Combination graph with motivation-to-combine and reasonable-expectation-of-success edges; secondary indicia weighted into the verdict.
Wands factors quantified — undue experimentation, working examples coverage, predictability of the art, breadth of the genus claimed.
Is the technical effect 'embraced and embodied' by the application as filed? Post-published evidence admissibility decided per jurisdiction.
Module 5 is a constrained optimiser. Variables are claim features. The objective is mean vulnerability, minimised across every interpretation × prior-art pair. Constraints: minimum scope, §112/Art. 83 clarity, jurisdictional consistency.
Like Gurobi, but for claims.
Classical: Gurobi · OR-Tools · SCIP for N ≤ 20 features. Heuristic: simulated annealing, genetic algorithms. Reinforcement learning with the validity score as the reward signal.
Excludes the bioisostere edge case that anticipated under DOE.
A pharmaceutical composition comprising at least one polymer P selected from P1, P2, P3, and at least one API X having an aqueous solubility ≤ 0.1 mg·mL⁻¹, wherein the composition exhibits a Tg of 50–100°C (±5%) and a dissolution rate of at least 20%·min⁻¹ at pH 6.8.
Adds a sub-genus disclaimer that survives Phillips and G 1/24.
[…] wherein the composition exhibits a Tg of 50–100°C, with a preferred sub-range of 60–85°C, excluding compositions wherein P is poly(ethylene-co-vinyl-acetate), and a dissolution rate of at least 20%·min⁻¹ at pH 6.8.
Inlines the technical effect to satisfy G 2/21 ab initio.
[…] wherein the composition exhibits a Tg of 50–100°C and a dissolution rate of at least 20%·min⁻¹ at pH 6.8, the said dissolution rate yielding a Cmax of API X ≥ 1.4× that of the unformulated reference, as embraced by the application as filed.
spaCy + LegalBERT/PatentBERT decompose the claim into atomic features F1, F2, … with quantifiers, ranges, and Markush groups.
Six-axis cartesian product. 10⁵–10⁸ configurations per claim, parallelised across the corpus.
Espacenet + USPTO + PATSTAT + PubChem + ChEMBL + Reaxys — full-text + structure + semantic search.
4-vector score per (interpretation × prior art): anticipation, obviousness, sufficiency, plausibility. Calibrated probabilities.
Constrained optimisation — minimise mean vulnerability, subject to scope and clarity constraints. Gurobi · OR-Tools · RL.
PDF executive summary + interactive dashboard + JSON for the deposit pipeline. 3–5 candidate claims, each with a robustness score.
CIL consumes IP Radar's landscape and feeds Doctrinal Hardening with optimised claims for screening.
The lattice ships in Q3 2026. Lock the Scale tier now to be on the launch list — or talk to a strategist about early access.