# AWS Blog Shows How to Plug Amazon S3 Vectors Into NVIDIA's NeMo Agent Toolkit as Agent Memory

> The AWS Machine Learning Blog published a walkthrough for using Amazon S3 Vectors as a custom persistent-memory backend in NVIDIA's open-source NeMo Agent Toolkit, deployed on Amazon EKS.

- **Topic**: Models
- **Published**: 2026-10-02T00:15:03.238Z
- **Canonical URL**: https://highsignal.sh/stories/aws-blog-shows-how-to-plug-amazon-s3-vectors-into-nvidias-nemo-agent-toolkit-ced89d69

## Why It Matters

Agent memory is becoming a build-it-yourself infrastructure decision: AWS is pitching S3 Vectors, which it says scales to 2 billion vectors per index with no capacity planning and pay-per-use pricing, as the storage layer under an open-source orchestration framework that already ships Mem0, Redis, and Zep backends. The post lands the same day attention is dominated by Gemini 4 Argon, a reminder that much of the practical AI work is plumbing, not frontier launches.

## Primary Sources & Citations

- [Build agent memory with NVIDIA NeMo Agent Toolkit and Amazon S3 Vectors](https://aws.amazon.com/blogs/machine-learning/build-agent-memory-with-nvidia-nemo-agent-toolkit-and-amazon-s3-vectors/) — *AWS Machine Learning Blog* (Primary source)

---

[← Back to Headlines](https://highsignal.sh/)