AMP: A Vendor-Neutral Wire Format for Agent Memory Operations

2026-05-31 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Software Development & Engineering · Depth: Expert, quick

Summary

memorywire is a new JSON-Schema 2020-12 wire format designed to standardize memory operations for agent-memory frameworks such as mem0, Letta/MemGPT, and Cognee. It addresses the current fragmentation where each framework uses proprietary SDKs and storage layouts, hindering integration and migration. The format defines five core memory operations—remember, recall, forget, merge, and expire—across four memory types: semantic, episodic, procedural, and emotional. It incorporates a MemoryStore interface, a fan-out router, and an optional Human-In-The-Loop governance channel for review. An open-source reference implementation supports five backend adapters, including sqlite-vec and pgvector. Performance benchmarks on a 100-fact / 50-query corpus achieved recall@5 = 1.000 for 42 queries, with ingest p50 at 37.8 ms and recall p50 at 40.6 ms. Adversarial-fusion experiments showed Reciprocal Rank Fusion maintained recall@5 = 1.000 under specific conditions, outperforming max fusion. A conformance suite passed 68 of 80 cells, validating its robustness as a vendor-neutral protocol.

Key takeaway

For AI Architects and Engineers designing or integrating agent memory systems, memorywire provides a critical vendor-neutral standard. Adopting this JSON-Schema 2020-12 wire format will significantly simplify your framework integrations and migrations, eliminating bespoke development for each memory solution. Furthermore, its optional Human-In-The-Loop governance channel offers a crucial mechanism for reviewing memory writes, enhancing the reliability and trustworthiness of your agent's long-term knowledge. Consider implementing memorywire to streamline your development and ensure robust memory management.

Key insights

memorywire standardizes agent memory operations, enabling vendor-neutral integration and governance across diverse frameworks.

Principles

• Agent memory requires a shared wire format.
• Governance is critical for long-term memory.
• Established components can form new protocols.

Method

memorywire defines a JSON-Schema 2020-12 wire format for five memory operations and four memory types. It integrates a MemoryStore interface, fan-out router, and optional HITL governance.

In practice

• Implement memorywire for agent interoperability.
• Use RRF for robust memory fusion.
• Integrate HITL for memory governance.

Topics

• Agent Memory
• Wire Format
• JSON Schema
• Interoperability
• Reciprocal Rank Fusion
• Human-In-The-Loop

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Scientist, AI Engineer, AI Architect

Related on AIssential

• Perplexity announces hybrid AI system that decides what runs locally or in the cloud — Perplexity's hybrid AI orchestrator intelligently routes tasks between local and cloud models for optimal privacy and efficiency.
• Codex Isn’t Just for Coders Anymore. OpenAI Is Coming for the Entire Office. — The expanding non-developer use of Codex is driving OpenAI's release of new enterprise features.
• Inside Google’s System for Coordinated A/B Testing Across Its Global Service Fleet — Google's system standardizes large-scale A/B testing across distributed services, ensuring statistical rigor and minimizing interference.
• Article: Two Misconfigurations That Caused Spark OOM Failures on Kubernetes — Cloud migrations can introduce silent infrastructure misconfigurations that cause Spark OOM failures under production load.
• You Can Finally Build Your Own LLM. Here’s Why You Probably Shouldn’t. — The build-versus-buy decision for LLMs is primarily an arithmetic problem, often favoring API rental over self-hosting.
• Introducing workspaces for Lambda Cloud — Lambda Cloud "workspaces" provide logical containers for GPU resources, enhancing access control and environment separation.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.