NexAgent AI
Vancouver local team online
NexAgent AI Consultant
Facing roadblocks? Our local Vancouver team provides 1v1 consulting for private AI deployments. Join our Discord community NexAgent Club or email [email protected].
tl;dr — NexAgent publishes two open-source projects: openclaw-nextagent (21 custom skills, pgvector memory, systemd) and Agentic Memory (PostgreSQL + pgvector + bge-m3). Both MIT licensed.
Building AI infrastructure in the open
Postgres + pgvector long-term memory plugin for OpenClaw — public release
Replaces OpenClaw's bundled SQLite memory with a Postgres 16 + pgvector backend designed around how human memory actually retrieves things. 4-tier recall (T0 in-process → T1 hot cache → T2 multi-route → T3 cold gist), multi-key Xinhua-dictionary indexing across 8 parallel routes, deterministic-first ingest (0 LLM tokens in the hot path), hard per-agent SQL-layer isolation, real-time bilingual dashboard. 75% of repeat queries return in <5ms. Apache 2.0.
Complete reference implementation: from official OpenClaw to production customization
21 custom skills across memory, content ops, credential vault, and multi-server ops. pgvector persistent memory. Discord-native interface. AES-256 encrypted secret refs — no plaintext credentials in code. Runs fully native (zero Docker) on systemd. nextagent.ca's blog + service-page automation is powered by this stack.
Earlier evolution — superseded by nextclaw above. Kept for reference.
Give your AI agent persistent, searchable, evolving memory. Powered by PostgreSQL + pgvector + bge-m3. Built as an OpenClaw plugin that fully replaces built-in SQLite memory with a unified PostgreSQL backend.
docker compose up -dpip install -r requirements.txtuvicorn main:app --port 18800curl http://localhost:18800/healthPublic release of nextclaw, a Postgres + pgvector long-term memory plugin for OpenClaw. 4-tier recall, multi-key Xinhua-dictionary indexing across 8 parallel routes, deterministic-first ingest, hard per-agent SQL-layer isolation, real-time bilingual dashboard. Apache 2.0 on GitHub.
Added agent-reach (17-platform web read: Twitter, Reddit, YouTube, Bilibili, XiaoHongShu, Douyin, Weibo, WeChat, Xueqiu, RSS, Exa...), server-inventory skill for multi-host ops over Tailscale SSH, and openclaw-updater self-upgrade flow. Skill count: 17 → 21.
AES-256 vault replaces plaintext refs for sudo password, SSH keys, Tavily keys, API tokens, and DB creds. Skills updated to pull secrets at runtime — no credentials in tracked code.
Full customization config released: 17 skill templates, database schema, systemd units, blog automation. Session memory fix: 2h idle reset + pre-compaction flush at 30k tokens.
All 5 Docker containers (PostgreSQL, Redis, MinIO, Flask API, Discord Bot) migrated to systemd. ~2GB freed, faster boot, zero Docker dependency.
Switched from local Qwen3-30B to Claude Sonnet 4.6. Integrated Tavily for news fetching. Built-in SQLite memory fully replaced via Agentic Memory plugin.
File permission fixes, secrets pulled out of Dockerfiles into .env, Discord allowlist enforced, security monitor bot deployed.
First open-source release of PostgreSQL + pgvector + bge-m3 memory system. Vector search, structured facts, workspace indexing, auto-extraction, time decay.
We help businesses deploy production-grade AI infrastructure. The same technology powering our open-source projects is available as a managed service.
Talk to UsWe open source two projects: openclaw-nextagent (a complete production AI agent configuration) and Agentic Memory (a PostgreSQL + pgvector memory plugin for OpenClaw).
Both projects are released under the MIT license — free to use, modify, and distribute in personal and commercial projects.
Open a GitHub issue or pull request on the respective repo. For major changes, discuss in our Discord first. We review contributions regularly.
Yes. MIT license permits commercial use. If you want a managed deployment or customization, NexAgent offers AI infrastructure services for that.