Ben Zanghi

AI-powered personality framework based on cognitive and behavioral patterns. V2 in development with enhanced assessment algorithms.

Memory system for conversational AI based on cognitive science principles (consolidation, decay, reconsolidation). 3,030 memories, 87.9% entity coverage.

Evolutionary search over AI agent patterns. Found that direct prompting beats multi-step workflows at 30B scale. 12 candidates × 10 generations × 5 benchmarks.

Interactive educational simulation visualizing AI capability progression from rule-based systems to transformers. Interactive timeline and model comparisons.

Two-tier voice assistant with local GLM routing and Claude deep path. 85% simple queries handled at 700ms latency. Smart home + semantic memory.

AI-powered personalized children's stories with voice cloning. Live product demonstrating generative AI for creative applications.

AI grocery shopping assistant with smart lists, price tracking, and personalized recommendations. Influenced NineTendencies architecture.

Personal AI paper monitoring with automated summarization and relevance scoring. Direct precursor to Engram memory patterns.

Building a production memory system for conversational AI based on cognitive science principles. Dual storage, temporal validity, uncertainty tracking, and entity deduplication in production.

Evolutionary search over AI agent orchestration patterns. 18 experiments across 6 models (7B–405B parameters) reveal that orchestration effectiveness depends entirely on model scale.

Building a production voice assistant with local GLM routing and semantic memory. 85% of queries handled at 700ms latency with smart home integration and Engram memory.