Aegis

Architecture

Aegis is built as the document-intelligence companion to ParaEval. Its job is to turn claims material into typed evidence, retrieve corroborating context, and produce a recommendation that stays transparent about uncertainty.

System Layers

Layer
Implementation
Purpose
Presentation
Next.js App Router pages + case workbench
Aegis is now surfaced as native routes inside the portfolio site rather than a separate Vite deployment. That keeps navigation, SEO, styling, and deployment unified.
Document intelligence
OCR + VLM parsing workflow
OCR and visual reasoning convert uploaded PDFs and screenshots into typed facts. The same processing service now supports both the standalone route and the case workflow.
Policy context
Typed policy framework + case metadata
Each case maps extracted facts to policy thresholds, permitted evidence types, and review notes so the recommendation stays grounded in contract context.
Evidence retrieval
Knowledge lookup over supporting artifacts
Aegis retrieves sensor logs, field reports, and modeled outputs that help an underwriter decide whether the uploaded document is sufficient, corroborated, or weak.
Decision handoff
Recommendation, risk flags, downstream evaluation
The immediate output is a recommendation, not a hidden final answer. When the case becomes trigger-sensitive, the file can hand off to ParaEval for stricter adjudication.

Current Integration Status

The source material began as a standalone Vite and Express app underaegis-document/. The portfolio now exposes the product surface natively through Next.js routes so it behaves like the rest of the site.

The current MVP cut is case-based: a user creates a case, uploads a document, runs extraction, and reviews the resulting evidence and recommendation in one place.

Live document parsing now sits behindapp/api/aegis/*route handlers, with the next increment being durable persistence rather than more demo scaffolding.

Guardrails

  • • Keep PII scrubbing explicit before wider retrieval or escalation.
  • • Separate document understanding from trigger adjudication so each step remains explainable.
  • • Expose evidence quality and missing data rather than pretending every upload is decisive.
  • • Gate live parsing and model calls behind environment configuration in production.

Relationship To ParaEval

Aegis and ParaEval should remain separate product surfaces. Aegis specializes in document understanding and triage. ParaEval specializes in formal trigger evaluation once structured evidence is available. Keeping those boundaries clear makes both systems easier to explain, test, and evolve.

Open Aegis casesView ParaEval