Concept
What is a portable application graph?
A portable application graph is the data structure that connects your applications, questions, answers, variants, and reviews through typed edges representing lineage, coverage, and fit. Instead of treating each application as an isolated document, the graph links related questions and answers across programs so that improving one answer benefits every application that touches the same underlying question. AQUA Application Hub is built on a portable application graph — it is the substrate that makes answer reuse, fit scoring, and lineage tracking possible.
Nodes in the graph
The application graph is made up of several types of nodes, each representing a distinct unit of application infrastructure:
- Applications. A specific submission to a specific program — the container that ties questions and answers to a destination.
- Questions. The underlying prompts a program asks, normalized so that the same question asked different ways maps to one node.
- Answers. Your strongest response to an underlying question, stored once and reused across programs.
- Variants. Program-specific adaptations of a source answer, tuned for framing, word limits, and evaluation criteria.
- Reviews. Feedback and review-loop records attached to answers or variants, preserving the history of how an answer evolved.
Edges in the graph
Nodes are connected by typed edges that carry meaning, not just references:
- Lineage edges. Connect a variant back to its source answer and supporting evidence, so you can trace why an answer says what it says.
- Coverage edges. Connect a program's question surface to the answers in your bank that address each question, powering fit scoring.
- Fit edges. Connect your profile to a program through computed fit signals, so the graph carries live scoring state, not just static content.
Why portability matters
Most application tools treat each submission as a standalone document. When you improve an answer for one program, that improvement is trapped inside that application. A portable application graph breaks that isolation: because answers, questions, and lineage are shared across programs, an improvement to a source answer propagates to every variant and application that depends on it. Portability also means the graph is not tied to a single program's format — it follows you across accelerators, fellowships, grants, and jobs, accumulating value as your application infrastructure grows.
How AQUA implements the application graph
AQUA stores the application graph as the core data model behind its answer bank, Smart Matcher, and scoring engine. Questions are normalized to underlying prompts so reuse is automatic. Answers are linked to variants through lineage edges, and coverage edges connect your answer bank to each program's question surface for fit computation. The graph is queryable through the web application and, for power users, through the local MCP server — so agents can retrieve answers, trace lineage, and write back improvements without leaving the graph model.