Agentic Plan Caching (APC)
https://neurips.cc/virtual/2025/poster/116166
Extracts, stores, adapts, and reuses structured plan templates from planning stages of agent applications. NeurIPS 2025 poster.
At a glance
- Type
- Test-time plan-template memory
- Tier
- T3
- Created
- 2025-06
- Latest release
- not applicable — not OSS
- License
- not applicable — not OSS
- GitHub
- not applicable — no GitHub repo
- Pricing
- not applicable — not commercial
- Funding
- not applicable — not commercial
Taxonomy
- storage
- kv
- retrieval
- similarity
- persistence
- cross-session
- update
- extraction
- unit
- skill
- governance
- inspectable
- conflict
- llm-arbitrate
When to use
Optimised for: not applicable - research paper
Anti-fit: not applicable - research paper
Pros & cons
Pros
Cache-aware planning reduces repeated work for similar tasks.
Cons
Niche use case; cache-invalidation issues common in evolving domains.
Claims & capabilities
Plan template extraction from completed agent runs with keyword retrieval and lightweight model adaptation. Headline: 50.31% serving-cost reduction and 27.28% latency reduction on average while preserving 96.61% of optimal performance; baselines: Accuracy-Optimal (large planner LM, no caching) and Semantic Caching; primary datasets: FinanceBench, TabMWP, QASPER, AIME 2024/2025, GAIA.
Technical surface
- API surface
- not applicable — research paper
- Backend storage
- not applicable — research paper
- Deployment
- not applicable — not a deployable product
- Embedding model
- not applicable — research paper
- Multi-tenancy
- not applicable — research paper
- MCP
- not applicable — research paper, no deployed product
- A2A
- not applicable — research paper, no deployed product
- OpenTelemetry
- not applicable — research paper, no deployed product
Similar systems
Other recent method papers — theorized, no distinct product in the catalog, ranked by inbound references.
- Compressive Transformer T3
Maintains recent states in full resolution while compressing older memories with learned compression functions. DeepMind.
- MemGPT v2 / agent-tools T3
Already in catalog as the foundational MemGPT paper. Note: Letta is the productionised successor (cross-listed).
- Transformer-XL T3
Extends context through segment-level recurrence + caching of hidden states from prior segments. Foundational long-context architecture.
- Generative Agents T3
Park et al. — landmark agent-simulation paper. Reflection + memory stream + retrieval enable believable agent behavior.
- MemoryBank T3
Enhances LLMs with long-term memory. Early influential paper.
- Reflexion T3
Language agents with verbal reinforcement learning.