Universal Academic Oracle | Guided Learning Platform

Universal Academic Oracle is an advanced educational ecosystem designed to maximize long-term retention and structured cognition rather than shallow answer dumping. Serving as a multi-mode intelligence framework, the platform is engineered around two distinct paths: Learning Mode and Exam Mode.

Instead of acting as a traditional transactional chatbot, the engine relies on a scientifically grounded discovery loop to guide users through progressive hinting, error correction, active recall, and rigorous structural evaluations.

Core Dual-Engine Architecture

1. Learning Mode (The Socratic Engine)

Optimizes for deep conceptual retention, pattern extraction, and durable intuition. Key learning systems include:

• Hint-Based Reasoning Flow: Delivers progressive scaffolds rather than instant solutions to encourage discovery.
• Feynman-Technique Reinforcement: Triggers when fundamental misunderstandings are identified to reconstruct knowledge from first principles.
• Context-Aware Follow-Up Pipeline: Implements selection-aware actions rendering interactive UI containers instead of raw text injections to keep learners in flow.
• Oracle Memory System: Centralizes tracked knowledge, specific mistakes, and continuous session updates across execution tracks.

2. Exam Mode (The Performance Track)

Transforms academic preparation by simulating real testing pressures and training independent decision-making under strict conditions. Key evaluation mechanisms include:

• Timed Simulation & Tiered Help: Supports configurable constraints extending from Level 0 (strict exam conditions with no hints) up to Level 3 (full worked solutions).
• Multi-Format Exam Intake: Processes structured input including PDF parsing, automated mark scheme ingestion, grouped exam section detection, and verbatim reading passage extraction.
• Examiner-Style AI Grading: Runs step-based marking logic matching curriculum standards (such as IGCSE, A-Level, AP, and SAT formats) to output grade boundaries and detailed diagnostics.
• Performance Memory Injection: Synthesizes full exam summaries, direct question prompts, and student corrections into the cognitive model.
• Blind-Checklist Synthesis: Combines cross-session statistics into personalized pre-exam guides before test days.

Integrated Support Modules

Integrated Quiz Platform

Generates automatic concept-specific quizzes directly linked to active Oracle Memory tracking data. Includes per-topic configuration caching in sessionStorage, loading locks to eliminate stale launches, and wrong-answer follow-up chip routing for targeted chat-level revisions.

Learning Dashboard

Surfaces visual learner summaries reflecting academic levels, circular learning-efficiency indicators, strengths, weaknesses, and direct action triggers to purge or focus specific tracked topics. Also permits complete session summaries to be downloaded for external reviews.

Intelligent Request Routing & Security Architecture

The engine operates on a secure distributed backend architecture utilizing production-grade Supabase Edge Functions. Underneath, an orchestration pipeline establishes strict parameter parameters:

• Gemini-First Orchestration Pipeline: Leverages advanced models (including Gemini 3.1-lite and related Google GenAI nodes) as primary intelligence.
• Failure Tracking & Real-Time Recovery: Monitors format conflicts and rate limits per provider to automatically skip unretriable nodes, triggering an automated valid-response race mode across fallbacks when necessary.
• Prompt Security Layer: Isolates core prompt constraints and orchestration logic inside backend functions. Deploys upfront jailbreak heuristics and content validation safeguards that selectively disable web search access during abnormal interactions.

Who It Is For

Universal Academic Oracle serves secondary, university, and independent self-learners attempting to master complex technical concepts, review critical examination pipelines, or track learning history across long-term evaluation loops.

The platform represents a closed-feedback approach to instructional design, blending technical pedagogical discipline with secure, distraction-free educational engineering.