{"slug": "a-local-first-memory-daemon-for-ai-agents-sqlite-onnx-zero-api-calls", "title": "A local-first memory daemon for AI agents: SQLite + ONNX, zero API calls", "summary": "A developer has released Awareness, a local-first memory daemon for AI agents that uses SQLite and ONNX embeddings to keep agent context on the user's machine, with zero API calls. The daemon achieves 95.6% recall@5 on LongMemEval_S on an M1 with 8GB RAM, and can optionally sync to cloud pgvector for team sharing.", "body_md": "Cloud memory tiers have a fundamental problem: your agent context - the most sensitive data you have - leaves your machine. I wanted memory that never does.\n\nSo Awareness runs a local-first daemon:\n\n```\nnpx @awareness.market/local start\n```\n\nNo account. No API key. No cloud. SQLite + ONNX embeddings on your machine, served to Claude Code, Cursor, or any MCP client on localhost.\n\n| Storage | SQLite, human-readable, yours forever |\n| Embeddings | all-MiniLM-L6-v2 as ONNX (23MB) - zero API calls |\n| Retrieval | hybrid BM25 + vector RRF - 1.7s per query on an M1 8GB |\n| Memory types | knowledge cards, bi-temporal facts, conflict detection |\n| Upgrade path | optional cloud sync to pgvector when you want team sharing |\n\nThree things a cloud memory tier can never give you:\n\n95.6% recall@5 on LongMemEval_S, on an M1 with 8GB RAM and zero LLM calls at retrieval. Two competitors edge us by ~1 point on recall, running hosted stacks. Full methodology with the tables we do not win:\n\n[https://awareness.market/benchmarks](https://awareness.market/benchmarks)\n\nLocal-first does not mean local-only. Burst inference, team-shared memories, and cross-device sync are real needs - the daemon upgrades to cloud (pgvector) with one command when you need it. The point is that local is the default, not the fallback.\n\nRepo and SDK: [https://github.com/everest-an/Awareness](https://github.com/everest-an/Awareness)\n\nWhat would make you switch your agent memory to local-first?", "url": "https://wpnews.pro/news/a-local-first-memory-daemon-for-ai-agents-sqlite-onnx-zero-api-calls", "canonical_source": "https://dev.to/everest_an/a-local-first-memory-daemon-for-ai-agents-sqlite-onnx-zero-api-calls-2ahk", "published_at": "2026-08-23 04:47:13+00:00", "updated_at": "2026-08-23 05:13:16.090069+00:00", "lang": "en", "topics": ["ai-agents", "ai-infrastructure", "developer-tools", "machine-learning"], "entities": ["Awareness", "SQLite", "ONNX", "Claude Code", "Cursor", "MCP", "LongMemEval_S", "pgvector"], "alternates": {"html": "https://wpnews.pro/news/a-local-first-memory-daemon-for-ai-agents-sqlite-onnx-zero-api-calls", "markdown": "https://wpnews.pro/news/a-local-first-memory-daemon-for-ai-agents-sqlite-onnx-zero-api-calls.md", "text": "https://wpnews.pro/news/a-local-first-memory-daemon-for-ai-agents-sqlite-onnx-zero-api-calls.txt", "jsonld": "https://wpnews.pro/news/a-local-first-memory-daemon-for-ai-agents-sqlite-onnx-zero-api-calls.jsonld"}}