← All docs

📝 Progressive Memory

Token-efficient memory that grows with your project. Automatic summarization keeps context windows lean.

The Problem

AI agents accumulate context over time. Without management, memory files grow unbounded — eventually exceeding context windows or burning tokens on stale information.

The Solution: Progressive Summarization

📅 Daily Logs

Raw daily notes capture everything:

memory/
├── 2026-07-04.md  # Full detail
├── 2026-07-03.md  # Full detail
├── 2026-07-02.md  # Summarized
└── 2026-06-*.md   # Archived

🧠 MEMORY.md

Curated highlights only:

  • • Key decisions and rationale
  • • User preferences
  • • Lessons learned
  • • Active project context

Consolidation Strategy

Age Treatment Token Cost
Today + Yesterday Full detail loaded ~2-5K tokens
This Week Key points only ~500 tokens
This Month Summaries in MEMORY.md ~200 tokens
Older Archived, searchable via vectors 0 (until recalled)

Weekly Review Workflow

  1. 1 Read through this week's memory/YYYY-MM-DD.md files
  2. 2 Extract significant decisions, lessons, or insights
  3. 3 Update MEMORY.md with distilled learnings
  4. 4 Remove outdated info from MEMORY.md
  5. 5 Archive old daily files (keep summaries)

Result: Flat Token Cost

With progressive memory, your agent's context cost stays roughly constant regardless of project duration. A 6-month project uses the same tokens as a 6-week project — only the depth of history changes.

← Back to Neo