The improve platform

The improve platform is built as a single-tenant, server-based platform. Every installation is dedicated to one organisation - isolated infrastructure, isolated data, no shared environment. The platform is designed around four principles: reproducibility, evidence capture, security, and validation-readiness, the non-negotiable requirements of early research and regulated drug development.

Versioned repository: what the improve platform stores and how

The versioned repository is the single system of truth for all scientific artifacts - datasets, model code, execution outputs, reports, and metadata. Every artifact is automatically versioned, searchable, and permanently linked to its lineage on check-in. Nothing is lost. Every state is recoverable.

The improve platform uses a three-layer storage architecture:

File content - stored in customer-controlled storage buckets, encrypted in transit.
Metadata and audit trail - stored in a governed database.
Search index - enabling full-text and structured search across all repository artifacts including datasets, scripts, model outputs, reports, and metadata.

Controlled execution: how the improve platform runs analyses

Every analysis step in the improve platform runs in a sandboxed execution environment managed by the runserver - the execution engine at the core of the improve platform. The runserver enforces execution isolation: users cannot interfere with each other’s executions, tool versions and environments are locked per run, and every execution is recorded in the audit trail with full environment capture.

Tool installations are packaged as containerised environments, giving complete control over software versions and dependencies. Users select from registered execution environments per run - different NONMEM versions, R versions, or Python environments can coexist within the same installation.

Administrators register and maintain execution environments, including legacy tool versions required for long-term reproducibility and regulatory inspection readiness. The environment used for any historical analysis can be preserved and re-executed years after its original completion.

The improve platform is tool-agnostic. Supported tools include NONMEM, Monolix, PsN, R, Python, SAS, MATLAB, and many others. Desktop tools like RStudio, MATLAB, Positron and others are also supported. Tool versions are registered by administrators and managed through the containerised execution framework, eliminating version drift between users and workstations.

The improve platform connects to the high-performance computing and cloud infrastructure IT already manages. Supported schedulers are SLURM and LSF. AWS Batch is accessible via AWS ParallelCluster. CPU allocation, memory limits, and queue selection are specified per run, giving IT visibility and control over compute consumption without over-provisioning fixed infrastructure.

Governance: how improve controls access and compliance

Fine-grained, role-based access controls are enforced at the platform level. What each user can see, modify, or approve is determined by the system. Every access event is recorded in the immutable audit trail.

Disabling a user at the identity provider level immediately restricts access. Account inactivity management and contractor expiry dates are supported via scripted controls.

Every action performed in the improve platform - code execution, file modification, model run, dataset delivery, approval decision, access event - is recorded automatically with named user attribution, precise timestamps, and full context. The record is immutable and append-only. Nothing can be altered, deleted, or obscured after it is written, by any user including administrators.

The audit trail is permanently accessible in both electronic and human-readable format and cannot be disabled by any operator. Every entry satisfies the electronic record requirements of 21 CFR Part 11 and EMA Annex 11.

Usability: how scientists reach improve

The improve platform is accessible through multiple interfaces. Scientists choose the one that fits their working environment. Every interface connects to the same central governed repository. The access point changes; the record does not. Interfaces to improve are described here.

Integrations: how improve connects to your enterprise

The improve platform is designed to integrate with the enterprise systems which are already in place.

Single sign-on through OIDC/OAuth2, LDAP, and Active Directory, inheriting the organisation’s existing MFA policy. No separate credential system required.

The improve platform connects to clinical trial management systems, laboratory information systems, and raw data repositories via API, preserving lineage between enterprise data sources and analytical outputs. Integrations are built to fit the specific systems and variants your organisation uses. Study metadata may be retrieved directly from governed sources.

Custom-developed applications communicate with the platform through the REST API. The CLI (command line interface) allows for automation.

Infrastructure: where improve runs and how it scales

The improve platform is built as a single-tenant platform - every installation is dedicated to one organisation. There is no shared infrastructure, no shared environment, and no possibility of data from different organisations residing in the same system.

The improve platform supports PostgreSQL and Amazon Aurora PostgreSQL as the primary database. The database stores all repository metadata, audit trail records, and execution history.

The improve platform manages simultaneous access from multiple users through database-level locking, proven at production scale in regulated environments.

Two feature releases are delivered per year, with patch releases as needed. scinteco undergoes annual ISO 9001 quality management audits.

The improve platform building blocks

The improve platform delivers its capabilities through two kinds of building blocks: components developed and maintained by scinteco, covering the complete analytical workflow from dataset request to submission output, and interfaces that connect the improve platform to the scientific tools and enterprise systems your teams already use.

improve components

Visualizes every model candidate, transformation step, and decision branch as a navigable graph from raw source data to final submission output. Every node is linked to the exact dataset, code version, execution environment, and outputs used to produce it.

Structures the complete dataset request-to-delivery process as a governed traceable pipeline, from formal specification through pipeline execution, QC, and versioned delivery. Every dataset is permanently linked to its source data.

Automates generation of model reports using predefined R Markdown or Quarto templates. For statistical programming activities, also automates lineage and Reviewer’s Guide directly from governed dataset metadata.

Shell-level access to all platform functions on Windows, Linux, and Mac. The primary integration point for automation tasks, SAS batch pipelines, and HPC job submission.

The graphical desktop client for Windows. Colour-coded execution status, built-in version control, and automatic extraction of key model parameters from run outputs.

Interfaces and integrations

Brings the complete improve environment into Visual Studio Code. Browse analysis trees, check in changes, run simulations, and navigate lineage without leaving the editor. improve studio will be available with improve version 5

Full improve repository access from within RStudio. Browse analysis trees, check in changes, and trigger runs. Live at launch.

Direct programmatic access to all improve platform functions, enabling custom integrations.

questions about layers, runserver or repository?

Talk to our engineers

The improve platform

Versioned repository: what the improve platform stores and how

Controlled execution: how the improve platform runs analyses

HPC and cloud compute integration

Governance: how improve controls access and compliance

Access control

Audit trail

Usability: how scientists reach improve

Integrations: how improve connects to your enterprise

Enterprise authentication

Enterprise systems

API and automation

Infrastructure: where improve runs and how it scales

Database

Scalability

Release and validation cadence

The improve platform building blocks

improve components

improve Analysis Graph

improve Data Order - dataset request and delivery

improVision - automated reporting

improve CLI - command-line access

improve Workbench

Interfaces and integrations

improve studio - the Visual Studio Code studio extension

TreeExplorer for RStudio

REST API