我們為軟體架構開發了類似Git的版本控制和情境感知AI

I’m the maker of ArchtSoft. We’ve been working on a problem we kept running into while building and reviewing systems:
Architectural decisions change over time, but the history, rationale, and context behind those changes usually get lost.

We built ArchtSoft to treat architecture as a first-class, versioned artifact, not just diagrams.

What it does:

1. Architecture version control
   Every change to an architecture is versioned. You can visually diff versions and see what changed, when, and why. The goal is Git-like clarity, but for system architecture rather than code.

2. Component-level architecture decision records (ADR)
   Instead of keeping ADRs in separate docs, each component stores its own decision context: trade-offs, constraints, assumptions, and rationale. This makes architectures explainable and reviewable over time.

3. Context-aware AI for manual architecture design
   This isn’t “AI generates diagrams.” Architects still design manually. The AI continuously reads the architecture context (industry, domain, functional, and non-functional requirements) and suggests things like component purpose, security considerations, and compliance implications while you design.

4. Architecture-to-code scaffolding
   From an architecture, we generate project scaffolding (folders + boilerplate) that can be downloaded as a ZIP and used as a starting point.

Why we built it:
We saw architecture break down not because of bad tech choices, but because:

Decisions weren’t documented where they were made

The context was lost when teams changed

Architecture drifted without a clear history

We’re trying to reduce that gap.

Link: https://archtsoft.com

I’d appreciate feedback from people who’ve dealt with architecture reviews, long-lived systems, or platform evolution. Happy to answer questions.