# I spent 6 weeks building AI memory that learns from what ships

- URL: https://runtimenotes.com/blog/why-does-my-ai-keep-forgetting/
- Published: 2026-01-23, updated 2026-02-17
- Tags: ai, memory, engineering, open-source
- Summary: Most AI memory systems optimize for storage. I built one that optimizes for outcomes.

Every AI coding session starts as a blank slate. Decisions don't carry over. Debugging context evaporates. Conventions get re-explained.

I was losing 10-15 minutes per session just rebuilding context.

So I studied 11 memory systems. Impressive engineering - but they share the same flaw:

**Good advice and bad advice sit side-by-side forever.**

They store memory. They don't learn.

---

## The Insight

Memory systems don't improve because they don't know what *worked*.

I built Memory Layer around one bet: retrieval quality comes from feedback, not from storing more text or building fancier graphs.

The obvious follow-up: who provides feedback?

Manual curation doesn't scale.

**Tasks do.**

When work ships, the memories used during that task get promoted. When work stalls, the memories that led there stop surfacing. No thumbs-up buttons. No "rate this response." Just outcomes.

![Good memory rises, bad memory sinks](https://runtimenotes.com/12.jpeg)

---

## Results

After 6 weeks of real use:

- Retrieval precision: 70% to 90%
- Debugging familiar issues: 15 min to 3 min
- Context re-explanation: ~50% reduction

---

## The Series

1. **[Why most AI memory systems don't learn](https://runtimenotes.com/blog/what-i-learned-from-11-ai-memory-systems/)**
   The landscape, the gap and why tasks turned out to be the missing signal.

2. **[Building memory that learns](https://runtimenotes.com/blog/building-memory-that-learns/)**
   5-signal retrieval, asymmetric scoring, SQLite tradeoffs, edge cases.

3. **[One memory store, every AI agent](https://runtimenotes.com/blog/one-memory-store-every-ai-agent/)**
   MCP, REST, SDK, Claude Code plugin, Beads and Claude Tasks integration.

4. **[Adding Runtime Memory to Hermes Agent as a memory provider](https://runtimenotes.com/blog/adding-runtime-memory-to-hermes-agent/)**
   The store, now called Runtime Memory, tested in one-shot Hermes Agent runs on nine coding tasks.

---

Memory Layer is open source: [github.com/runtimenoteslabs/memory-layer](https://github.com/runtimenoteslabs/memory-layer)

```bash
pip install git+https://github.com/runtimenoteslabs/memory-layer.git
mem add "Always validate JWT server-side" -c convention
mem search "authentication"
```

One store. Every agent. Learns from what ships.