ORIGINAL COMPARISON · PRIMARY SOURCES ATTACHED

Three AI architecture patterns, compared.

JARVIS, Trading Swarm, and Indie Metrics MCP solve different problems, but each uses an explicit boundary to make an AI system easier to inspect and replace.

THESIS

The shared design idea is controlled composition.

JARVIS controls how experience components connect. Trading Swarm controls how independent signals become consensus. Indie Metrics MCP controls which operations an assistant can perform. Their strongest common pattern is not autonomy. It is an explicit contract around autonomy.

COMPARISON MATRIX

Different problem. Different boundary.

01

Pluggable experience orchestration

JARVIS

How can an AI desktop experience stay modular?

Components
Scene, voice, brain, script, feeds
Risk boundary
Privacy changes with the selected adapters
Language
TypeScript
02

Adaptive multi-signal consensus

Trading Swarm

How can several imperfect agents reach one decision?

Components
Six technical signals plus sentiment
Risk boundary
Historical tests do not establish future performance
Language
Python
03

Constrained read-only tool mediation

Indie Metrics MCP

How can an assistant query sensitive business data safely?

Components
Revenue, customer, product, subscription, transaction, refund, forecast
Risk boundary
Keys and code still require careful review
Language
TypeScript

METHOD

DESCRIPTIVE, NOT A BENCHMARK

How this comparison was produced.

The matrix was extracted from the public README and architecture documentation for each repository on July 27, 2026. It compares documented system boundaries and does not rank code quality, adoption, security, profitability, or production readiness.