Agent Workflow

1 · Ingest an evidence dataset

JSON (schema v1.2) or CSV. Drop a file, or load the bundled synthetic SLX-402 dataset.

⤒

Drop evidence file here

or click to browse · .json / .csv

Load synthetic dataset (24 items) ⇩ Sample JSON ⇩ CSV template ⇩ Rubric JSON

2 · Orchestration settings

Controls how the agent team executes.

Human-in-the-loop on Critical

Pause the pipeline for reviewer sign-off on every Critical item

Reviewer

Recorded in the audit trail

Agent pacing

Slow it down to watch agents collaborate

FastNormalDemo

Primary market for deck

Drives headline timing in the summary

3 · Multi-agent pipeline

Agents in the same column run concurrently; the Risk Rubric agent gates on human review for Critical items.

Ingest→Verify ∥ Organize→Risk + HITL→Gaps ∥ Escalate→Summary ∥ Charts→Analytics

Agent message bus

Clear

Live log of inter-agent messages and decisions.

Run snapshot

Updated live as agents complete.

Evidence Library

Organized by the Evidence Organizer agent. Click any row for the full record, verified citation and rubric score breakdown.

⇩ Export CSV

All domains All markets

AllLowMinimumHighCritical

ID	Evidence	Domain	Type	N	Key result	GRADE	Score	Tier	Citation

Evidence Risk Rubric

Every evidence item is scored on seven transparent factors. Any critical trigger overrides the score and forces human-in-the-loop review and escalation.

Scoring factors

Points are summed; tier assigned by total.

Critical triggers & recommended actions

Auto-detected by the Risk Rubric agent, or declared in the dataset.

Methods references

Published standards grounding the rubric (real citations).

Gap Analysis

Coverage of each evidence domain against the requirements of six HTA bodies. Cells show the best available evidence tier; planned and excluded items do not count as coverage.

▦

Run the agent pipeline to generate the gap matrix.

Domain × HTA agency coverage

R = required · E = expected · O = optional · — = not assessed by agency. Hover a cell for supporting evidence.

CoveredPartial (High-risk only)At risk (Critical only)Gap (no usable evidence)

Prioritised evidence gaps

P1 = required evidence missing or at risk; P2 = required partial / expected missing; P3 = expected partial.

Market readiness

Weighted coverage (R×3, E×2, O×1).

Executive Evidence Summary

Drafted by the Executive Summary agent. Text is editable — your edits flow straight into the PowerPoint export.

↻ Regenerate⧉ Copy text⇩ Export to PowerPoint

¶

Run the agent pipeline to generate the executive evidence summary.

Charts

Produced by the Visualization agent. Hover for detail; download any chart as PNG or SVG.

▥

Run the agent pipeline to render charts.

Evidence portfolio by domain and risk tier

Number of evidence items

Relative treatment effects

Hazard / odds / risk ratios with 95% intervals, log scale · <1 favours solvextinib for HR

Evidence readouts vs HTA submission calendar

Data cut-off dates by domain; dashed lines = planned submissions

HTA market readiness

Weighted evidence coverage by agency (%)

Rubric score by evidence item

Tier bands shaded; ◆ marks items with a critical trigger

Top-line Analytics

Maintained by the Analytics agent — a standing scorecard of the evidence base, refreshed every run and after every human decision.

⇩ Analytics JSON

◔

Run the agent pipeline to populate analytics.

Tier distribution

Share of active evidence items

Evidence by publication status

Items

Evidence hierarchy

Level assigned by Organizer agent

Domain scorecard

Items, mean rubric score, pivotal share

Run history

Every pipeline execution this session

Critical Escalations

Generated by the Escalation agent for every item confirmed Critical. Recipients: cbankslead@gmail.com, milesambrosebanks@gmail.com.

✉ Open digest e-mail⇩ Download all (.eml)

⚑

No escalations yet. Critical items confirmed during human review will appear here.

Audit Trail

Every human-in-the-loop decision and agent milestone, time-stamped for inspection readiness.

⇩ Export audit JSON

Time	Actor	Event	Item	Detail