Built to connect, not to replace
The improve platform is designed to connect - with the scientific tools scientists already use, with enterprise infrastructure IT already manages, and with the AI assistants that are increasingly part of how scientists work. This page covers all current and forthcoming integrations.
Scientific tool integrations
The improve platform is tool-agnostic. It does not replace existing scientific tools but wraps around them, adding versioning, reproducibility, evidence capture, and governed execution to the analyses scientists already run.
The deepest and most established integration in the improve platform. NONMEM runs are submitted through improve’s execution layer, with automatic extraction of key parameters - OFV, Thetas, Omegas, Sigmas - from run outputs. Every run is versioned, traceable, and captured with its complete execution environment. improve by scinteco’s original user base is built on NONMEM pharmacometricians.
The improve platform integrates deeply with R through the improveR package, enabling child-step rerun, lineage navigation, and run submission directly from R code. TreeExplorer brings the improve repository into RStudio - browse analysis trees, check in changes, and trigger runs without leaving the IDE. (Coming soon) R Shiny applications can be built, versioned, and shared with system and non-system users through improve.
Python can be used with local installations and is also supported with the improve CLI and with containerised execution environments. Every Python run is versioned and captured with its complete dependency environment. Documentation and tooling specifically for Python-first scientists is under active development.
SAS batch pipelines integrate with the improve platform through the CLI, enabling governed execution of SAS dataset generation and transformation scripts. Dataset generation scripts in Data Order are SAS-based, with automated variable replacement driving extraction from raw data repositories.
MATLAB integrations are supported through the improve CLI, designed specifically for long-running tool integrations where interactive session management is not practical.
All supported as registered execution environments within the improve platform’s containerised framework. Runs submitted through any of these tools are versioned, traceable, and captured with their complete execution context.
The Uppsala University Pharmpy toolchain, developed in collaboration with Roche, is integrated as an improVerticle, enabling fully automated model development with complete traceability inside the improve platform. PharmR is supported alongside Pharmpy as part of the automated modeling pipeline.
IDE integrations
improve studio brings the full improve environment into Visual Studio Code - browsing analysis trees, checking in changes, triggering runs, and navigating lineage without leaving the editor. Full parity with the improve Workbench.
In a standard Visual Studio Code setup, scientists can install any extension from the public marketplace - including unvalidated tools that fall outside IT’s governance framework. In regulated pharma, this creates a compliance risk that is difficult to manage by policy alone.
The improve platform will integrate with Coder, a platform that runs Visual Studio Code as a remote development environment hosted on IT-controlled infrastructure. Within a Coder-managed environment, IT determines exactly which extensions are available - scientists can install only what has been approved and validated. improve studio is pre-approved and centrally deployed; no unvalidated software can be introduced alongside it.
The result is the full IDE experience scientists expect, within a governed environment IT can stand behind.
Full improve repository access from within RStudio - browse analysis trees, check in changes, and trigger runs directly from the IDE scientists working in R already use.
Enterprise system integrations
The improve platform connects to the organisation’s existing identity providers through OIDC/OAuth2 and LDAP. Single sign-on, MFA inheritance, and centralised access management - no separate credential system required.
The improve platform connects to clinical trial management and laboratory information systems via API. Integrations are built to fit the specific systems and variants your organisation uses, preserving lineage between enterprise data sources and analytical outputs.
The Data Order module connects to raw data repositories including entimICE, enabling governed, automated dataset retrieval with full lineage from source data to modeling output.
The improve platform submits runs to HPC environments through SLURM and LSF. Resource usage, CPU allocation, memory limits, queue selection - all specifiable per run, giving IT visibility and control over compute consumption.
The improve platform can run on AWS infrastructure - S3, EC2, Aurora PostgreSQL, Route 53, and OpenSearch. AWS Batch is accessible using AWS ParallelCluster.
AI integrations
The improve platform exposes an MCP server that AI assistants - including Claude, GitHub Copilot, Cursor, and any MCP-compatible tool - can connect to from any IDE or language environment. Through this integration, AI can operate inside improve as a governed participant with its own bounded permissions, and its own audit trail entries.
Critically, the AI has the same tools as a human user - it can browse the repository, submit runs, check in changes, and navigate lineage - but only within its assigned permissions, and everything it does is fully traceable. This means organisations can give scientists access to AI assistance within the governed improve environment without losing the audit trail, the reproducibility, or the regulatory defensibility they need to keep. As the EU AI Act begins to mandate audit trails over AI-assisted decisions, the MCP integration provides a governed path to AI adoption in regulated M&S activities.
Custom integrations with internal systems, data sources, and automation pipelines are supported via the improve REST API. scinteco provides integration support on a time-and-materials basis.