Stop guessing whether your practice is healthy.
Plumbline gives you a structured, repeatable way to assess the state of your architecture practice — and the tools to improve it. Assess, author, conform.
A readiness score that reflects reality
An assessment runs against your published architecture principles, grouped into sections — Business Architecture, IS Architecture, Technology Architecture, Security, and more. Plumbline scores each section from 0 to 100 based on how completely your principles cover the relevant concerns.
The score is not a vanity metric. It reflects gaps in coverage that would leave your teams without clear architectural guidance. Each low-scoring item links directly to the principle gap driving it.
- ✓Section-by-section breakdown with gap detail
- ✓Historical trend — see whether your practice is improving
- ✓Linked directly to the principles driving each score
AI-drafted principles you actually review
When an assessment detects a coverage gap, the AI-author generates a principle suggestion to address it. Suggestions land in an inbox — not automatically published, not silently applied. You review each one, accept it as-is, edit it to your organisation's voice, or reject it with a note.
The architect stays in control. The AI handles the blank-page problem.
- ✓One suggestion per detected gap, not bulk noise
- ✓Accept, edit, or reject — full audit trail on every action
- ✓Accepted suggestions go directly into draft principle workflow
Conform: continuous principle enforcement
The Conform module — coming after the current Assess + Author release — will watch your landscape and automatically flag elements that violate published principles. Drift from principles surfaces as Findings, routed to Reviews when needed.
Plumbline then covers the full loop: assess the state of your practice, author the principles that address the gaps, and enforce them against the live architecture.
Bring your own LLM key
The AI-author module uses Anthropic's Claude API. You configure your Anthropic API key at the tenant level — no key shared between tenants, no Northwise markup on your AI spend. You pay Anthropic directly at their published rates.
This also means you choose the model. Claude Haiku for speed and low cost, Claude Sonnet or Opus when quality of output matters more. Your preference, your budget.