Executable mathematics
Native models, SBML and COMBINE/OMEX packages, simulation code and a pinned execution environment. The EHMN2027 scaffold is supplied separately.
Models you can audit and trust.
Public-knowledge and proprietary foundation models for model-informed drug development: executable models, graded evidence and a configured LLM layer, delivered as one versioned package.
PRODUCT OVERVIEW 01 / 09Original presentation · open or download the complete PDF.
The delivered foundation model brings together the LLM layer and the files that make its outputs traceable. A locally configured AI PC is optional.
Native models, SBML and COMBINE/OMEX packages, simulation code and a pinned execution environment. The EHMN2027 scaffold is supplied separately.
19 organised folders spanning evidence ledgers, calibration, verification, validation, analyses, reports, regulatory cross-references and risk registers.
Configured prompts, exact run history, checksums, retrieval over project files and 22 critic checks before an evidence-backed claim is advanced.
Cloud-assisted, hybrid or fully local workflows. An optional AI PC can ship with an open-weight LLM and the IQANOVA runner configured and tested.
Timelines are targets from the agreed brief; the written quote defines exact scope, model families, licensing and price.
Start with public science. No confidential data required.
Add your own protected drug and experimental information.
Both tiers are scoped deliveries; an estimated duration is not a guaranteed completion date. Local-model performance is benchmarked against frontier output before reliance, as described in the overview.
Source provenance, units and rate laws, calibration and validation evidence, run history, uncertainty, applicability and explicit claim boundaries.
Datasets are distinguished as real, semi-synthetic or synthetic, with the origin of variables identified. Literature evidence is graded from A1 to E.
Examples from the November 2026 overview; each actual project needs its own context-of-use and evidence assessment.
For a T-cell engager class: what MABEL-based starting-dose range follows from published data, and which parameters drive it?
Is an approved drug's labelled dose adjustment consistent with a mechanistic exposure prediction?
Do rate laws, units and assumptions withstand review, and which proposed claims go beyond the supporting evidence?
Choose cloud-assisted public-data analysis, a hybrid mode or fully local operation for sensitive work. Every release has an evidence cut-off date, manifest and change log.
Evidence Watch, Model Maintenance and portfolio updates are optional services that start only under an activated agreement.
See deployment options in PDF ↗Send your assets, model family, questions, stage and context of use. We will respond with a written scope and price.
Research use and decision support only; not medical, regulatory, legal or investment advice. All services, delivery dates and prices depend on agreed terms.