📘 RayRay for Dummies

The friendly guide to not breaking things while helping build a PDF extraction system.

What Even Is This Thing?

RayRay is a tool that helps defense analysts extract data from PDFs about radar systems (and other detection stuff) without losing track of where the numbers came from.

The big idea: Click any extracted value → see exactly where it came from in the original PDF. No more "where did you get that number?" in meetings.

Important rule: AI does the first pass, but a human MUST review before anything gets saved. This isn't optional - it's required by OMB Memo M-25-21.

Quick reference for every page in RayRay and what it's for.

📁

Documents

/documents

"Your PDF library"

Upload, manage, and organize PDFs and spreadsheets. This is where the extraction workflow starts.

Who:Analysts
Features:
Upload PDFs/ExcelView document listStart extractionsComparison history tab
📄

Document Detail

/documents/[id]

"Deep dive into one document"

View a single document, run extractions, review AI results, and see page-by-page source highlighting.

Who:Analysts, Reviewers
Features:
Page viewerExtraction panelSource highlightingApprove/Reject/Modify
âš–ī¸

Compare

/compare

"Side-by-side analysis"

Select multiple documents and compare their extracted values. Scores and ranks them based on a scoring profile.

Who:Analysts, Managers
Features:
Multi-document comparisonScoring profilesRanked resultsExport comparison table
📡

Radar

/radar

"Radar system database"

Browse and search radar systems. View extracted specs, compare capabilities, and see data sources.

Who:Analysts, Reviewers
Features:
Radar searchCapability cardsSource verificationComparison integration
đŸ“ģ

Sources

/sources

"Intel feed control center"

Manage data ingestion pipelines. Monitor OSINT feeds, official sources, research papers, and internal reports.

Who:Analysts, Admins
Features:
Source status monitoringActivity feedEnable/disable sourcesReliability grades
âš™ī¸

Settings

/settings

"Your account & preferences"

Manage your profile, security settings (including MFA), and developer tools.

Who:Everyone
Features:
Profile settingsMFA setupDeveloper toolsSession info
📈

Scoring Profiles

/settings/profiles

"Define how scoring works"

Create and manage scoring profiles used in comparisons. Set weights, normalization methods, and attributes.

Who:Analysts, Managers
Features:
Create profilesSet attribute weightsNormalization methodsDefault profile selection
🏠

Home

/home

"Your command center"

The landing page after login. Shows quick stats, recent activity, and navigation shortcuts.

Who:Everyone
Features:
Recent documentsActive comparisonsQuick actions
📊

Dashboard

/dashboard

"The big picture"

High-level overview of system activity, stats, and metrics. Good for managers and quick status checks.

Who:Managers, Reviewers
Features:
System statsActivity chartsTime savings estimates
📜

History

/history

"Your audit trail"

View past extraction workflows, reviews, and comparisons. Essential for M-25-21 compliance.

Who:Reviewers, Managers
Features:
Workflow historyReview recordsAudit trailExport logs
đŸ—ēī¸

Roadmap

/roadmap

"Where we're headed"

Development roadmap showing completed, in-progress, and planned features.

Who:Everyone
Features:
Feature statusRelease timelineUpcoming features
🔨

Builder's Log

/builders-log

"Dev changelog"

Developer notes on what's been built, changed, or fixed. Technical details and decisions.

Who:Developers, Curious users
Features:
Build historyTechnical notesArchitecture decisions
🍮

The Pudding Page

/the-pudding-page

"The proof is in the pudding"

Hard numbers on ROI, time savings, and why RayRay matters. Great for convincing leadership.

Who:Managers, Decision makers
Features:
Time savings statsROI calculationsBefore/after comparisonsFTE impact
â„šī¸

About

/about

"What is RayRay?"

Overview of the platform, mission, and team. Good for onboarding new users.

Who:New users
Features:
Platform overviewMission statementTeam info

Common Oopsies & Fixes

Reading the Logs

The API terminal shows what's happening. Here's the decoder ring:

[UPLOAD]PDF being uploaded
[WORKFLOW]Extraction process steps
[LLM ANTHROPIC]AI is thinking
[ERROR]Something broke
[WORKFLOW ERROR]Extraction specifically broke

The Golden Rules

  1. 1Always review extractions - M-25-21 requires it
  2. 2Start API before Web - Order matters
  3. 3Check your API keys - If extraction fails, 90% of the time it's the key
  4. 4Click-to-source is your friend - Use it to verify
  5. 5When in doubt, restart both terminals - Clears a lot of weird state

Still stuck? Check the API logs in the uvicorn terminal.

Questions? Check the logs first, then ask a developer.