claude-mem vs TurboOCR

claude-mem is an open-source memory compression plugin that gives Claude Code a persistent brain across sessions. It hooks into six Claude Code lifecycle events to automatically capture tool observations, compress them into semantic summaries, and store everything in a local SQLite + Chroma vector database. When a new session starts, relevant context is injected automatically — no copy-pasting, no re-explaining architecture decisions you made last week. The system achieves roughly a 10x token reduction through progressive disclosure: it retrieves only what's relevant for the current task rather than dumping everything into context. Developers can query their memory store via natural language through MCP tools (search, timeline, get_observations), and a built-in web viewer at localhost:37777 lets you inspect memory streams visually. Privacy controls via <private> tags let you keep sensitive content out of the store. Install is a single npx command, and it works with Claude Code, Gemini CLI, and OpenClaw gateways. The project hit 48K+ GitHub stars and is clearly scratching a real itch: the loss of context between sessions is one of the most consistent pain points for AI-assisted development.

TurboOCR is a C++20 OCR server that uses CUDA and TensorRT to process documents at speeds that make Python-based OCR look like a fax machine. The headline number: 270 images per second on FUNSD form datasets with approximately 11ms single-request latency — roughly 50x faster than PaddleOCR's standard Python implementation. It uses PP-OCRv5 models (the same underlying tech as PaddleOCR) but squeezes them through TensorRT FP16 optimization for GPU inference. The server exposes both HTTP and gRPC interfaces from a single binary and handles PDFs natively with four extraction strategies: pure OCR, native text layer extraction, hybrid verification mode, and a "best of both" fallback chain. PP-DocLayoutV3 handles layout detection across 25 document region classes — useful for structured documents where you need to know that a bounding box is a table cell vs. a header vs. a figure caption. A Prometheus metrics endpoint tracks throughput, latency, and GPU memory in real time. Deployment is Docker-first: TensorRT engine compilation happens automatically on first startup. The catch is it requires Linux with an NVIDIA Turing GPU (RTX 20-series minimum) and driver 595+, so it's not a laptop tool. But for enterprise document automation — invoices, forms, medical records — the throughput-to-cost ratio is hard to beat.