# AI-assisted development lifecycle (AI DLC)

**NR-004** · Delivery framework · v1.0.0 · Reference / illustrative

Show where AI accelerates software delivery and where independent engineering evidence remains essential.

![AI-assisted development lifecycle (AI DLC)](diagram.svg)

## Architecture problem
Fast code generation can increase changes faster than teams can validate behavior, design integrity, security and operations.

## Apply when
- Designing engineering standards for teams using AI-assisted coding
- Defining approval and evidence requirements per delivery stage
- Explaining how AI is used throughout the lifecycle, not only coding

## Not intended for
- Treating generation as validation
- Letting agents approve their own outputs without independent evidence
- Requiring AI at every step regardless of suitability

## Design rules
1. Human accountability and architectural decision ownership are explicit.
2. AI can assist at every stage but does not remove verification gates.
3. Test design and security controls should be created before or alongside implementation.
4. Production signals feed requirements, evaluation and architecture change.
5. Release promotion requires independently verifiable evidence.

## Trade-offs
- Stricter gates can slow local iteration while reducing production regressions.
- Automated evaluations need maintenance and can drift as business behavior evolves.
- Autonomous coding helps throughput only when test and deployment systems are reliable.

## Review questions
- What can the agent change and with what repository permissions?
- Which tests are independent of generated implementation?
- Who accepts an ADR or a business-rule exception?
- What signals trigger rollback and learning?

## Download / adapt
- `diagram.drawio` — primary editable diagrams.net file
- `diagram.dot` — independent Graphviz concept source
- `diagram.mmd` — editable Mermaid interpretation
- `diagram.svg` — scalable figure
- `diagram.png` — high-resolution figure
- `metadata.yaml` — structured pattern description

**Reuse:** CC BY 4.0; credit nishtnr, link the license, and indicate changes. Conceptual reference only.