We don't add AI to software. We design software around AI.
The AI-First Company develops methodologies, frameworks, and tooling for teams who believe that AI is not a feature — it is a foundational architectural layer.
Our work sits at the intersection of software craftsmanship and agentic AI: clean architecture, domain-driven design, and rigorous engineering discipline applied to a world where the developer, the QA, the product owner, and the AI agent collaborate as peers.
We publish what we consider ready to be used, forked, and challenged. Everything else is work in progress — or proprietary.
| Repository | Description |
|---|---|
saaster-kit |
Production-grade starter kit — Flutter · Rust · Kong · Temporal · SigNoz — pre-wired for AI-assisted development |
flutter-firebase-starter |
Flutter + Firebase starter kit — Clean Architecture · BDD · flutter_bloc |
forge |
AI development tooling — the best of BMAD, SpecKit and Agent OS, unified into a single opinionated workflow |
sai-code |
Agentic AI Coding CLI tool with your own models |
An extension of Event Modeling (Adam Dymitruk) with a fifth swimlane dedicated to AI agents.
Three interaction patterns: AI-Command, AI-Enriched View, AI-Automation.
Designed to make AI agency explicit, traceable, and architecturally intentional — not bolted on.
A BDD session format that brings together Product Owner · Developer · QA · LLM as equal participants.
The LLM is not a tool that assists the session. It is a collaborator that challenges acceptance criteria, surfaces edge cases, and stress-tests assumptions in real time.
A development workflow built around machine-readable specifications as the primary artifact.
Specifications drive code generation, test generation, and agent orchestration — not documentation.
SDD is the operating layer that connects AI Modeling outputs to agentic tooling like Claude Code.
SatisfactorIA — a conversational interface for AI-First project specification.
Helps product teams articulate what they want to build, expose what they have not thought through, and produce structured, agent-ready specifications.
Part of a forthcoming SaaS suite. No timeline communicated until it is ready.
- We do not ship AI wrappers around existing tools and call it transformation.
- We do not believe prompt engineering alone is an engineering discipline.
- We do not treat developer productivity metrics as a proxy for software quality.
- We do not publish until we have used it ourselves in production contexts.
Public repositories accept issues and pull requests.
We review seriously and respond honestly — including when the answer is "no".
For partnership, consulting, or speaking engagements, reach out via the contact information on our public profiles.
Crafted with discipline. Designed to last.