Agentic System as Compressor: Quantifying System Intelligence in Bits

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Zihan Qin, Hongrui Zhang
• Venue: arXiv: cs.AI
• Published: 2026-06-24
• Collected: 2026-06-25
• URL: http://arxiv.org/abs/2606.25960v1
• PDF: https://arxiv.org/pdf/2606.25960v1
• DOI: UNKNOWN

Plain-Language Brief

Researchers propose a method to quantify the intelligence of an agentic system, which includes tools, evidence retrieval, environment constraints, verifiers, and multi-turn interactions, by measuring how many bits are needed to reconstruct a target object under a fixed task distribution, interface, and compute budget. The study demonstrates this approach in five settings and finds that using agentic components reduces the number of bits required.

What the Document Says

• The focus is on large language models evolving from isolated predictors to agentic systems that call tools, retrieve evidence, obey environment constraints, use verifiers, and complete tasks through search and multi-turn interaction.
• The authors adopt the viewpoint that 'compression is intelligence': a stronger agentic system allows for a target object to be reconstructed with fewer bits under a fixed task distribution, interface, and compute budget.
• Three methods are operationalized for measuring codelength: arithmetic coding, seed coding, and a fallback.
• The study evaluates this measure in five settings: reversed text, chess moves, protein sequences, retrieval-augmented question answering, and semantic story compression.
• In all the settings, agentic components reduce codelength.

Key Concepts Mentioned

• large language models
• agentic system
• compression is intelligence
• codelength
• arithmetic coding
• seed coding

Claims Made by the Document

• Agentic systems can be quantified by measuring the number of bits needed to reconstruct a target object.
• The evaluated methods for codelength reduction (arithmetic coding, seed coding) work in all five settings.

Evidence Used

• In the study, the authors demonstrate the codelength reduction with agentic components in various settings.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Large language models are turning from isolated predictors into agentic systems: they call tools, retrieve evidence, obey environment constraints, use verifiers, and complete tasks through search and multi-turn interaction. We adopts an analytical viewpoint based on 'compression is intelligence': under a fixed task distribution, interface, and compute budget, a stronger agentic system lets a target object be reconstructed with fewer bits. We operationalize the measure with arithmetic coding, seed coding, and a fallback, and evaluate it in five settings: reversed text, chess moves, protein sequences, retrieval-augmented question answering, and semantic story compression; in all of them agentic components reduce codelength. These small, controlled experiments cover component types typical of real agentic systems, show that codelength can analyze how components, observers, and budgets change residual uncertainty, and offer guidance for evaluating real agent systems.