Loading…
Loading…
Vibe coding generates code from prompts. Spec-driven development generates code from a specification — your architecture, data model, and compliance requirements, encoded once and compiled into production-ready code across 9 stacks.
SOC 2 · ISO 27001 · GDPR · HIPAA · PCI-DSS · DORA · NIS2 — generated from spec flags
In February 2026, Andrej Karpathy wrote that the vibe coding era is ending. The category delivered on speed-to-prototype. It did not deliver on production readiness — no architecture docs, no compliance controls, no migration history, non-deterministic output.
Spec-driven development is the next phase. The specification is the source of truth, not the prompt. Archiet compiles your PRD into an Architectural Genome — a structured YAML model of your system's entities, workflows, compliance requirements, and technology choices. Code is derived from the spec, not approximated from language.
The practical difference: when a compliance requirement changes, you update one flag in the Genome and regenerate. Every affected file updates consistently. With vibe coding, you re-prompt across multiple sessions and manually reconcile the differences.
The Genome is version-controlled in git. Your architecture lives next to your code, not in a chat log that gets deleted.
Upload your PRD, paste a description, or fill in the Architecture Wizard. Archiet extracts entities, workflows, compliance requirements, and technology choices.
Your system is encoded as an Architectural Genome — a structured YAML spec you can inspect, edit, and version-control. This is your specification, not a prompt log.
Choose from 9 backend stacks or Tauri+Rust for desktop. The same Genome generates across all stacks — change once, regenerate everywhere.
The ZIP contains backend, frontend, mobile app, database migrations, CI/CD, Docker, compliance artifacts, and architecture documentation. Zero-touch production-ready.
Different tools for different stages. Vibe coding wins for demos and exploration. Spec-driven wins for production, compliance, and long-term maintenance.
| Capability | Spec-driven (Archiet) | Vibe coding (Bolt, Lovable, v0) |
|---|---|---|
| Source of truth | Formal specification (Architectural Genome) | Prompt history (conversation) |
| Output reproducibility | Deterministic — same spec always produces same output | Non-deterministic — re-prompting gives different results |
| Compliance controls | Generated from spec flags (HIPAA, SOC 2, DORA, GDPR) | Prompt-dependent — usually omitted |
| Database migrations | Full Alembic / Prisma migrations auto-generated | Schema only — no migration history |
| Authentication | JWT httpOnly cookies, RBAC, forgot-password, email verify | Basic auth — varies by prompt quality |
| Multi-tenancy | Workspace isolation enforced at every query | Not included |
| Architecture documentation | ArchiMate model generated and version-controlled alongside code | None — no record of design decisions |
| CI/CD + Docker | GitHub Actions, Dockerfile, docker-compose, K8s manifests | Not included |
| When requirements change | Update spec → regenerate — all files update consistently | Re-prompt and reconcile manually |
| Best for | Production apps, regulated industries, teams maintaining code for years | Hackathons, demos, rapid prototyping |
Spec-driven development starts with a formal specification — an architectural model that encodes your data entities, workflows, compliance requirements, and technology choices — and generates production code from that specification deterministically. Unlike prompt-driven generation (vibe coding), the output is derived from the spec, not approximated from natural language. When the spec changes, regeneration is consistent and predictable.
Vibe coding tools (Bolt, Lovable, v0) generate code from natural language prompts. They are fast for exploration and prototyping. Spec-driven development generates code from a structured specification. The result is reproducible, compliance-aware, and architecturally consistent. Vibe coding is the right tool for demos; spec-driven is the right tool for production.
The Architectural Genome is Archiet's specification format. It is a YAML document that describes your system's modules, fields, compliance flags, vendor integrations, workflows, and technology choices. It is machine-readable, version-controllable with git, and diff-able. The Genome is the spec that drives all code generation — backend, frontend, mobile, database, CI/CD, and compliance artifacts.
Flask + Next.js, FastAPI, Django + Next.js, NestJS + Next.js, Laravel + Next.js, Go chi, Java Spring Boot, Rails + Next.js, .NET Core, and Tauri + Rust (desktop). All from the same Architectural Genome.
Yes. When your Genome includes compliance_flags, Archiet generates the corresponding compliance artifacts alongside the code. Supported: SOC 2, ISO 27001, GDPR, HIPAA (Security Rule Technical Safeguards), PCI-DSS, DORA (EU 2022/2554), NIS2. Compliance controls are structural — enforced in the generated code, not added as documentation.
Yes. Import your existing architecture document or diagram into the Architecture Wizard. Archiet extracts the entities, services, and compliance requirements and produces the Genome. You can also describe your system in plain English and have the Genome generated from your PRD.