# hingel-augment-how-ai-changes-sdlc-six-stages-2026-06-08

## Veille

Guide by Augment Code (Paula Hingel) describing how AI agents are restructuring the software development lifecycle (SDLC), stage by stage. Thesis: AI produces **higher throughput at some stages and higher instability risk at others** — a symptom of uneven adoption without redrawing review boundaries. Draws on **DORA 2025**: AI adoption correlates positively with delivery throughput and product performance, but **negatively with stability**. Six stages revisited (Requirements, Design/Architecture, Implementation, Testing/QA, Deployment, Maintenance), three major risks (erosion of the junior pipeline, **circular validation** of AI-generated tests, governance gaps at scale) and three emerging roles (**Intent Engineering**, Agentic DevOps, AI Governance/Assurance). Actionable recommendations: audit one stage before scaling, stress-test governance, make the **specification** central, define explicit rollback policies, redesign the junior role around review.

## Titre Article

How AI Changes the SDLC: A Six-Stage Guide

## Date

2026-06-08

## URL

https://www.augmentcode.com/guides/how-ai-changes-the-sdlc

## Keywords

SDLC, software development lifecycle, coding agents, specification, specification-driven development, circular validation, vibe architecting, governance, DORA 2025, delivery throughput, stability, change failure rate, rollback, intent engineering, agentic DevOps, AI governance, junior pipeline, review, orchestration, complexity debt, blast radius

## Authors

Paula Hingel (Augment Code)

## Ton

Profile: long-form corporate blog guide (Augment Code), expert third-person perspective, analytical and prescriptive register, high technical level backed by research sources (DORA 2025, CMU study, SWE-bench Pro, Meta DRS), target audience of engineering leaders, architects, EMs, platform/DevEx leaders, and governance decision-makers. The tone is that of a **structuring framework**: no hype, but a stage-by-stage reading of the SDLC that systematically names the central tension (throughput vs. stability) and addresses it through redrawn review boundaries. The rhetoric alternates between **quantified diagnosis** (DORA correlations, +30% static-analysis issues, +40% complexity, 70% of developer time spent on code comprehension) and **operational prescription** (audit one stage, stress-test governance, rollback policies). Striking signal neologisms: "vibe architecting" (architectural decisions made in seconds without governance) and "circular validation" (AI generates both the code and the tests confirming it). Authority rests on the combination of a proven product (CLI Augment, top SWE-bench Pro score at 51.80%) and a cited research corpus; the commercial character stays understated, with the article presenting itself first as a governance reference for the agentic SDLC.

## Pense-betes

- **Central tension**: AI creates "higher throughput in some stages and higher stability risk in others" — the consequence of **uneven adoption without redrawing review boundaries**.
- **DORA 2025 anchor**: AI adoption has a **positive relationship with delivery throughput and product performance**, but a **negative relationship with delivery stability**. → the foundation (process, review) matters more than the tool.
- **Six-stage model** (agent + human): 1. **Requirements & Planning**: the **specification becomes the control mechanism** that directs the agent; humans focus on requirement quality and resolving ambiguity. 2. **Design & Architecture**: more architectural decisions require **explicit human review** to avoid **"vibe architecting"** (choices made in seconds without governance). 3. **Implementation**: the developer shifts toward **orchestration, validation, approval** rather than writing code. 4. **Testing & QA**: QA focuses on spec quality and coverage judgment; the core risk = **circular validation** (AI tests confirm AI code instead of verifying the actual requirement). 5. **Deployment**: throughput gains create **stability risks** → need for **stronger rollback controls**. 6. **Maintenance & Operations**: agents handle detection and remediation; humans manage **exceptions and hardening**.
- **Three major risks**: (1) **erosion of the junior pipeline** (automating foundational tasks faster than junior roles are redesigned shrinks the future pool of seniors); (2) **circular validation** in testing; (3) **governance gaps at scale** (oversight capacity becomes the limiting factor).
- **Three emerging roles**: **Intent Engineering** (translating ambiguous business objectives into testable specs), **Agentic DevOps/Infra** (deploying/training/orchestrating agents — LangGraph, Crew AI, Autogen), **AI Governance/Assurance** (overseeing multi-agent outputs, defining accountability boundaries). Hiring signals: Accenture, Scale AI, OpenAI ("minimum 1 year with LLMs", "agentic framework" experience).
- **Cited research figures**: **SWE-bench Pro** — CLI Augment **51.80%** (Feb. 2026, top published score at the time); **Meta DRS** — >10,000 changes landed during a code freeze, minimal production impact (2024); **code comprehension = ~70% of developer time** (major bottleneck); **CMU study of 807 GitHub repositories** — static-analysis issues **+~30%**, complexity **+>40%**; **DORA** — openly addressing job-displacement fears = **+125%** team AI adoption, dedicated learning time = **+131%**.
- **5 recommendations**: (1) **audit one stage** of the SDLC before scaling agents (clarify autonomy vs. human gates); (2) **stress-test existing governance** (architecture reviews, release controls, ownership); (3) make **specification governance** central; (4) **explicit rollback policies**; (5) **redesign the junior role** around review/validation without losing learning opportunities.
- **Related**: converges with Atlassian's AI-native SDLC (impact measurement), Williams' ADLC series (deterministic gates, tests-as-spec against reward hacking — a direct echo of circular validation), Rafal ("the faster execution goes, the stricter the framework must be"), Pragdave *Failing Faster* (complexity debt), Dropbox/Okumura (bottleneck displacement).

## RésuméDe400mots

This guide from Augment Code, authored by Paula Hingel, proposes a **six-stage** model for understanding how AI agents are restructuring the software development lifecycle. Its central thesis: AI does not uniformly improve the SDLC — it increases throughput at some stages while raising instability risk at others. This imbalance is not a technological inevitability but the symptom of uneven adoption carried out **without redrawing review boundaries**. The article draws on the **DORA 2025 report**, which establishes a positive correlation between AI adoption and throughput/product performance, but a **negative correlation with delivery stability**: process maturity matters more than the tool.

The six stages are reread through this lens. (1) **Requirements & Planning**: the specification becomes the control mechanism that directs the agent; humans focus on requirement quality and resolving ambiguity. (2) **Design & Architecture**: more decisions require explicit human review, to avoid "vibe architecting" — infrastructure or integration choices made in seconds, faster than governance can keep pace. (3) **Implementation**: the developer shifts from writing code to orchestration, validation, and approval. (4) **Testing & QA**: the core risk is **circular validation**, where AI-generated tests confirm AI-generated code instead of verifying the actual requirement; a precise specification is the safeguard. (5) **Deployment**: throughput gains create stability risks, hence the need for stronger rollback controls. (6) **Maintenance & Operations**: agents handle detection and remediation, humans manage exceptions and hardening.

Three structural risks are named: **erosion of the junior pipeline** (automating foundational tasks faster than junior roles are redesigned shrinks the future pool of seniors), circular validation, and governance gaps at scale. In parallel, three roles emerge: **Intent Engineering** (translating ambiguous objectives into testable specs), Agentic DevOps/Infra (orchestrating agents), and AI Governance/Assurance.

The guide is backed by data: 70% of developer time spent understanding existing code, a CMU study (807 repositories) showing +30% static-analysis issues and +40% complexity, and Meta's DRS system (>10,000 changes landed during a code freeze). It closes with five operational recommendations: audit one stage before scaling, stress-test governance, make the specification central, define explicit rollback policies, and redesign the junior role around review.

## GrapheDeConnaissance

- Augment Code —publie→ How AI Changes the SDLC: A Six-Stage Guide (DOCUMENT, 0.96)
- Paula Hingel —a_créé→ How AI Changes the SDLC: A Six-Stage Guide (DOCUMENT, 0.93)
- agents IA —améliore→ débit des étapes (CONCEPT, 0.9)
- SDLC AI-native —est_variante_de→ SDLC (METHODOLOGIE, 0.9)
- agents IA —affirme_que→ l'adoption inégale accroît le risque d'instabilité sans frontières de revue redessinées (AFFIRMATION, 0.9)
- DORA 2025 —mesure→ adoption IA positivement liée au débit mais négativement à la stabilité de livraison (MESURE, 0.92)
- specification-driven development —réduit→ validation circulaire (CONCEPT, 0.89)
- revue humaine d'architecture —résout→ vibe architecting (CONCEPT, 0.87)
- validation circulaire —s_oppose_à→ vérification du besoin réel par les tests (CONCEPT, 0.86)
- CLI Augment —mesure→ 51,80 % sur SWE-bench Pro (fév. 2026) (MESURE, 0.88)
- étude CMU sur 807 dépôts —affirme_que→ les issues d'analyse statique montent ~30 % et la complexité >40 % (AFFIRMATION, 0.85)
- intent engineering —s_applique_à→ traduction d'objectifs métier ambigus en specs testables (CONCEPT, 0.86)
- Paula Hingel —recommande→ auditer une étape du SDLC avant de scaler les agents (AFFIRMATION, 0.88)
- Paula Hingel —recommande→ définir des politiques de rollback explicites au déploiement (AFFIRMATION, 0.86)
- Paula Hingel —recommande→ redessiner le rôle des juniors autour de la revue et de la validation (AFFIRMATION, 0.86)
- Meta DRS —permet→ landing de >10 000 changements pendant un code freeze (2024) (MESURE, 0.84)

---
Canonical: https://www.thekb.eu/en/fiches/hingel-augment-how-ai-changes-sdlc-six-stages-2026-06-08/