Risk Triage

AI-powered agentic risk assessment tool with real-time reasoning, evidence citations, and exportable reports.

Risk Triage

Requires a Pro subscription.

Updated

pro
Open Risk Triage

Overview

Risk Triage is a multi-step agentic risk assessment tool that produces a comprehensive risk report for any company. Enter a ticker or company name, choose a depth preset (Quick, Standard, or Deep), and watch as the AI agent streams its reasoning in real time before delivering a full report with evidence citations, risk scores, and visual diagrams.

The tool supports advanced model presets for power users, optional focus instructions to hone in on specific risk areas, and parallel deep research for faster results. Every run is persisted locally (up to 50), and each has a unique shareable URL so you can bookmark or share specific assessments with colleagues.

How it works

Configuration

Enter a ticker or company name, choose a depth preset, and optionally focus the analysis.

AAPL
More
Focus on specific areas... (optional)
Run Risk Triage

Risk Levels

The completed report is assigned one of four risk levels based on the evidence found.

CRITICAL
Immediate attention required
ELEVATED
Significant risk signals found
NORMAL
Standard risk profile
LOW
Minimal risk indicators

Run Statuses

Queued
Analyzing
Complete
Failed

Analysis Pipeline

The agent executes up to 8 steps, streaming reasoning events in real time. The number of iterations (research loops) depends on the depth preset.

Initial Research
Gather filings and public data
Entity Extraction
Identify people, companies, relationships
Plan Research
Design targeted queries
Execute Research
Run queries across sources
Evaluate Evidence
Score confidence, find contradictions
Quality Check
Verify findings, fill gaps
Generate Report
Compile final risk assessment
Enhance Visuals
Add diagrams and charts

What you can do

Setup & Configuration

  • Search by Ticker or Company Name: optional Jurisdiction for name-based lookups
  • Depth presets: Quick (~2 min), Standard (~5 min), Deep (~10 min) — plus advanced model presets (GPT 5.1, Gemini, DeepSeek, Ollama, Mixed Fast, Local First)
  • Focus instructions: tell the agent to concentrate on specific risk areas
  • Advanced options: iteration count (0–10), visual enhancements, simple mode, entity skipping, no-cache, parallel deep research with processor selector

Analysis & Output

  • Real-time SSE event stream: agent reasoning steps shown as they happen
  • Completed report: rich Markdown with evidence links, source citations, and Mermaid diagrams
  • Risk level badge: CRITICAL / ELEVATED / NORMAL / LOW with colour coding

History & Sharing

  • Run history sidebar: up to 50 past runs persisted locally, each with subject name, date, and live status
  • Export report as PDF: download button opens print dialog
  • Shareable URLs: run ID in the URL (?runId=xxx) for bookmarking or sharing a specific run

Tips

  • ▸Start with "Quick" to get a rapid overview, then re-run with "Deep" if the initial results warrant it
  • ▸Use the "Focus on specific areas" field to hone in — e.g. "accounting irregularities" or "insider selling patterns"
  • ▸Watch the Agent Thinking panel during analysis to understand the AI's reasoning chain
  • ▸Click evidence citations in the report to see the underlying source documents
  • ▸The sidebar preserves your 50 most recent runs — use it to compare risk assessments over time
  • ▸Share a run with colleagues by copying the page URL — the runId parameter makes it persistent and bookmarkable

Related Pages

Red Flags

Aggregated AI-detected red flags across all tracked companies, with severity scoring, live market data, and manual submission.

Filings

Unified filings browser with category tabs, enriched columns, deep filtering, insider charting, and real-time updates.

Borrow Data

Securities lending data showing borrow fees, share availability, and hard-to-borrow indicators.