Skip to content
N
Booting Engineering OS
000%
All projects
Building#6

System Design Assistant

Your AI co-architect — interactive system design, diagrams, ADRs, and tradeoff analysis grounded in your codebase.

An AI co-architect that designs with you — diagrams, ADRs, and tradeoffs, grounded in your real code.

View repoAI co-architect
Status
Engine working: Mermaid diagrams, ADR drafting, weighted tradeoff analysis, design memory, interview grading. Apache-2.0.
Difficulty
Medium
Time to MVP
4–6 weeks

Role in the platform

A grounded module. Reuses Codebase Intelligence (#2) for grounding; ships as a ContextOS module plus a standalone interview-prep product.

Overview

An assistant that helps engineers design systems: generate and iterate on architecture diagrams (Mermaid/diagram-as-code), produce ADRs, reason about tradeoffs (scaling, consistency, cost), and — uniquely — ground designs in the actual codebase via Codebase Intelligence (#2).

Plus a prosumer interview-prep mode: practice system-design interviews with an AI interviewer/coach.

The problem

System design is high-stakes, under-tooled, and AI-assistable. Generic chatbots give unstructured advice with no persistence; diagram tools don't reason.

Grounding design in real code, not generic patterns, is the differentiator.

What it is

Pose a design problem; optionally ground it in your codebase (#2), reason through tradeoffs and options, and emit a Mermaid diagram + an ADR + a tradeoff analysis — then persist it as design memory you can iterate on.

Two distinct revenue pools — enterprise architecture and interview prep — de-risk it.

Key features

Diagram generation

Generate and iterate on architecture diagrams as Mermaid / diagram-as-code.

ADR drafting

Produce Architecture Decision Records that become the team's design record.

Tradeoff analysis

Weighted reasoning about scaling, consistency, and cost — not generic advice.

Codebase-grounded

Ground designs in your real architecture via Codebase Intelligence (#2).

Interview-prep mode

Practice system-design interviews with an AI interviewer/coach that grades answers.

Tech stack

TypeScriptMermaid / diagram-as-codeADR generationCodebase Intelligence (#2)Persistent design memoryLLM reasoning

Why it's different

  • Codebase-grounded design, not generic patterns.
  • Structured, persistent output (diagrams + ADRs + tradeoffs) versus a chat transcript.
  • Two markets de-risk it: enterprise architecture and interview prep.

Roadmap

MVP

Conversational design, Mermaid diagram generation, ADR drafting, tradeoff analysis, save/iterate.

V1

Codebase grounding (#2), design memory, teams, export (Confluence/MD), interview-prep mode.

V2

Multi-agent design review, cost/scaling estimators, ContextOS integration.

V3

Enterprise architecture governance, standards, on-prem.