Operations

Smartsheet keeps itself current from scheduled SAS/R — both the AI-assisted version and the no-code / no-AI version. Code owns the data; Smartsheet owns the alerts.

Outlook emails become a searchable knowledge base by dragging them into a synced folder — one paste-able PowerShell script. No Power Automate designer, no SharePoint clicking, no Copilot Studio.

The study KB as a navigable linked-note graph. Now a runnable R Shiny app (igraph metrics + a visNetwork force-directed graph you filter and click), a visual guide, a shareable interactive graph (self-contained, no server), and a build-it walkthrough — the three CSVs are the vault. Plus the 28-note vault and the design note.

The agentic trial-operations layer on the hybrid stack (recurrent QC, SOP RAG, eTMF filing, RBQM), plus the on-device small-model stack and the Copilot-free Study Operations KB.

Deterministic, no-AI monitoring that detects, packages, and pages humans — freshness gate → checks → roll-up → digest → tier-2 alert → heartbeat. The coverage playbook started with "covering while on leave" (service account, named backup, dead-man's switch) and now handles five scenarios: planned leave, unplanned absence, role handoff or departure, onboarding a new owner, and bus-factor review. The Continuity Toolkit turns a team's job inventory into a coverage matrix, a risk register, per-job checklists and a printable handoff packet, all in the browser.

The safety-monitoring backbone made visible: the eDISH Hy's-Law screen, QTcF (ICH E14), the 3+3 DLT ladder, AE by SOC×grade, and a per-participant evidence packet. Now a runnable R Shiny app (click the eDISH scatter to open a packet), a visual guide (the AND-gate as a diagram + live screens), a shareable dashboard, and a build-it walkthrough — all from one shared R logic core. A screening lens, never a reported number; the medical monitor / SRC / DSMB decide.

The 3-tier (participant / study / client) trial-termination early-warning console: signals roll up to a tier RAG and an overall risk index, with an advanced engine (Optimal Transport + Active Inference). Now a runnable R Shiny app (set a signal's level and watch the tier risk move), a visual guide (the roll-up as a diagram + a live console), a shareable dashboard, and a build-it walkthrough — all from one shared R logic core. The AI flags; the SRC / DSMB / PM / account lead decide.

The operational study-lifecycle console — portfolio RAG heatmap, ranked early-warning signals, milestone track, data readiness, TLF funnel, deliverables pipeline. Now a runnable R Shiny app you edit and own (with a transparent what-if on the risk score), a visual guide (the scoring model as a diagram + a live roll-up), a shareable dashboard, and a build-it walkthrough — all from one shared R logic core. Plus the operating-depth field guide. Synthetic, no PHI.

Run a UAT of the IRT/RTSM (randomization & supply) system before go-live, from the biostatistics lens — confirming the vendor build implements our randomization (allocation, stratification, blocking, concealment, dose-escalation logic), dispensing, resupply, and emergency unblinding. Delivered five ways from one shared R logic core: a visual guide (Mermaid diagrams + a live gate computed from the data), a runnable R Shiny app you edit and own, a shareable dashboard, a fillable Excel workbook (formula-driven go-live gate) for the eTMF, and a build-it walkthrough. Synthetic test data.

When an out-of-scope deliverable suddenly needs doing, where do the hours come from? A planned-vs-actual-vs-forecast (EAC) effort reconciliation: net findable = (Σ planned + contingency) − Σ EAC, then a decision tree — reallocate under-run slack → contingency → defer → change order. Delivered five ways from one shared R logic core: a visual guide (decision-tree + source-map diagrams, a live roll-up), a runnable R Shiny app, a shareable dashboard, a fillable Excel workbook, and a build-it walkthrough. Hours/effort only, no dollars; synthetic.