█████╗ ██╗██████╗ ██████╗ ██╗ ██████╗
██╔══██╗██║██╔══██╗██╔══██╗██║ ██╔════╝
███████║██║██║ ██║██║ ██║██║ ██║
██╔══██║██║██║ ██║██║ ██║██║ ██║
██║ ██║██║██████╔╝██████╔╝███████╗╚██████╗
╚═╝ ╚═╝╚═╝╚═════╝ ╚═════╝ ╚══════╝ ╚═════╝
The open standard for teams building AI-assisted software in regulated and enterprise environments.
"Humans define intent and validate outcomes. AI executes craft and maintains continuity. Every decision is preserved. Every output is auditable. Quality is a system property, not a human heroic act."
AIDDLC is an open specification that defines how teams should structure software development when AI participates in the development process itself. It answers the question AWS, Google, and every AI tooling vendor has avoided: not which tools to use, but how to work.
It is tooling-agnostic. It works with Claude, GPT, Gemini, any IDE, any cloud. Its value is in the structure and the principles — two organisations using completely different AI stacks should produce recognisably AIDDLC-compliant processes.
AIDDLC is authored and maintained by 10QBIT Technologies. The specification is open for community contribution under the governance model defined in GOVERNANCE.md. 10QBIT retains architectural authority.
AIDDLC Standard v1.0
├── Engineering Track ─────────────────────────────────────────
│ How AI-assisted teams build software
│
│ Foundation → Discovery → Architecture → Specification
│ → Build → Validation → Deploy & Learn
│
│ 7 phases · gate-controlled · compliance-aware by design
│ Proven in production: Club Health OS (healthcare SaaS)
│
└── Product Track ──────────────────────────────────────────────
How AI products find market fit
Discover → Prototype → Deploy → Scale
Model selection · AI-native metrics · feedback loops
Community development area — contributions invited
| # | Principle | What it means |
|---|---|---|
| 01 | Intent over implementation | Humans own the what and why. AI owns the how. |
| 02 | Context never dies | Every decision feeds every future decision. Zero decay at handoffs. |
| 03 | Gates before generation | No AI output without human-defined acceptance criteria. |
| 04 | Continuous validation | Quality verified at every phase, not tested at the end. |
| 05 | Auditable by design | Every output, decision, and rationale logged and attributable. |
| Repository | Purpose |
|---|---|
aiddlc/standard |
The canonical specification. Versioned releases. Issue queue = contribution channel. |
aiddlc/reference-portal |
Open-source reference implementation. The working AIDDLC-compliant project interface. |
aiddlc/examples |
Worked implementations by industry vertical. Healthcare first. |
aiddlc/certification |
Certification criteria, self-assessment checklist, certified organisation register. |
aiddlc/website |
Source for aiddlc.ai. Contributions accepted. |
AIDDLC artifacts are designed to satisfy the evidential requirements of regulated industries without additional instrumentation.
GDPR MHRA CQC GPhC FCA HIPAA ISO 27001 EU AI Act
→ Full regulatory mapping: aiddlc.ai/compliance
Use the standard → Read the specification at aiddlc.ai/spec
Implement it → Clone the reference portal at aiddlc/reference-portal
Contribute → Read CONTRIBUTING.md — every substantive proposal receives a written response
Get certified → aiddlc.ai/certification — founding cohort open
Enterprise support → standard@aiddlc.ai
Authored and governed by 10QBIT Technologies · Open contribution, architectural authority retained by 10QBIT · Standard v1.0 · 2026