# More context made memory worse. A smaller rule survived. — Galdo

> I gave memory extraction more prior context. It improved three LongMemEval answers and made fourteen worse. Here is what failed, what survived, and the one deterministic rule I shipped.

Canonical URL: https://getgaldo.com/blog/longitudinal-memory-consistency-experiment

I gave memory extraction more prior context. It improved three LongMemEval answers and made fourteen worse. Here is what failed, what survived, and the one deterministic rule I shipped.

## About Galdo

Galdo shows you how your company actually works. It learns how your best people work, as the work happens, and keeps the decisions, exceptions, and lessons behind the work, including how they change. People ask in Slack; agents ask through the API, SDKs, or MCP, and answers come back with the source. US-hosted, with a free tier of 1K writes and 10K searches per month.

## Explore Galdo

- [Galdo API documentation](https://getgaldo.com/docs)
- [Use cases](https://getgaldo.com/use-cases)
- [Benchmarks](https://getgaldo.com/benchmarks)
- [Company Brain for Slack](https://getgaldo.com/brain)
- [Ask Galdo widget](https://getgaldo.com/widget)
- [People — one memory container per person](https://getgaldo.com/people)
- [Framework integrations](https://getgaldo.com/integrations)
- [Trust and security](https://getgaldo.com/trust)
- [Contact](https://getgaldo.com/contact)
- [llms.txt](https://getgaldo.com/llms.txt)
- [OpenAPI 3.1 specification](https://getgaldo.com/openapi.json)

---

This is the Markdown representation of https://getgaldo.com/blog/longitudinal-memory-consistency-experiment, served by content negotiation on `Accept: text/markdown`. Machine-readable index: https://getgaldo.com/llms.txt · https://getgaldo.com/openapi.json · https://getgaldo.com/sitemap.xml