HN Showcase
Back to browse
GitHub Repository

Zero-cost local memory for AI assistants. Claude Opus + OpenClaw + nomic-embed + sqlite-vec. Full Telegram history → semantic search in 2.4s.

3 starsPython

Local memory for AI assistants – zero-cost Telegram history search

by tituss-bit·Feb 28, 2026·2 points·0 comments
Visit ProjectView on HN

AI Analysis

●●SolidSolve My ProblemNiche GemZero to One

Local semantic search for Telegram memory, but OpenClaw adoption and embedding quality unclear.

Strengths
  • •Eliminates cloud embedding costs ($5–10/month) with CPU-fast nomic-embed indexing (2.4s for 157 chunks).
  • •Bilingual benchmark validates nomic-embed-text v1.5 over larger competitors on conversational data.
  • •Git-based multi-machine sync avoids third-party memory storage, privacy-native design.
Weaknesses
  • •Tightly coupled to OpenClaw; generic AI assistants need config adaptation beyond documented scope.
  • •No live demo, no adoption signal; unclear if multilingual accuracy translates to English-only use cases.
Target Audience

AI assistant developers, privacy-conscious users integrating long-term context into chatbots

Similar To

Perplexity AI context indexing · Pinecone (self-hosted) · Milvus

Similar Projects

AI/ML●●Solid

OpenContext – Bring Your Own Coding Agent, Local-First, No Vendor Lock

The MCP server to give Claude true persistent memory is the clearest win — it tackles the annoying, real-world problem of lost context across sessions. The repo pairs local analysis (Ollama) with import/export for ChatGPT/Claude/Gemini and ships as Node/TypeScript + Docker with a usable web UI, so you can run everything without vendor APIs; I'd like to see wider model/import support and clearer UX around merges/conflicts, but the core idea and execution feel immediately useful.

Niche GemSolve My Problem
akarnam37
103mo ago