feat: init the article - vibecoding
This commit is contained in:
parent
bdb9fab97e
commit
2ff21717a8
|
|
@ -0,0 +1,46 @@
|
|||
---
|
||||
slug: ai-coding-methodologies-beyond-vibe-coding
|
||||
title: "AI Coding Methodologies: Beyond Vibe Coding"
|
||||
author: henry-technical
|
||||
status: inbox
|
||||
created: 2025-01-22
|
||||
updated: 2025-01-22
|
||||
content_type: explainer
|
||||
primary_keyword: ""
|
||||
secondary_keywords: []
|
||||
assets_folder: assets/ai-coding-methodologies-beyond-vibe-coding/
|
||||
---
|
||||
|
||||
# Idea
|
||||
|
||||
**Source:** Perplexity research on AI-assisted development terminology (Jan 2025)
|
||||
|
||||
**Concept:** Overview article covering AI coding methodologies landscape. Position vibe coding as casual/risky approach, contrast with professional techniques (Spec-Driven Development, AI-DLC, Agentic Coding, etc.).
|
||||
|
||||
**Goal:** Establish Henry's expertise in AI-assisted development space. Warm-up article for Dev.to account.
|
||||
|
||||
**Angle:** "Wide view" — Henry surveys the field and shares his practical preferences.
|
||||
|
||||
---
|
||||
|
||||
# Brief
|
||||
|
||||
*pending @strategist*
|
||||
|
||||
---
|
||||
|
||||
# Outline
|
||||
|
||||
See [outline.md](assets/ai-coding-methodologies-beyond-vibe-coding/outline.md)
|
||||
|
||||
# Draft
|
||||
|
||||
See [text.md](assets/ai-coding-methodologies-beyond-vibe-coding/text.md)
|
||||
|
||||
# SEO
|
||||
|
||||
See [seo-metadata.md](assets/ai-coding-methodologies-beyond-vibe-coding/seo-metadata.md)
|
||||
|
||||
# Activity Log
|
||||
|
||||
See [log-chat.md](assets/ai-coding-methodologies-beyond-vibe-coding/log-chat.md)
|
||||
|
|
@ -0,0 +1,55 @@
|
|||
---
|
||||
slug: beyond-vibe-coding
|
||||
title: "Beyond Vibe Coding: Professional AI Development Methodologies"
|
||||
author: henry-technical
|
||||
status: inbox
|
||||
created: 2026-01-22
|
||||
updated: 2026-01-22
|
||||
content_type: explainer
|
||||
primary_keyword: ""
|
||||
secondary_keywords: []
|
||||
assets_folder: assets/beyond-vibe-coding/
|
||||
---
|
||||
|
||||
# Idea
|
||||
|
||||
**Source:** Perplexity research on AI-assisted development terminology (Jan 2026)
|
||||
|
||||
**Concept:** Overview article covering AI coding methodologies landscape. Position vibe coding (Collins Word of Year 2025) as entry point, then survey professional approaches: Spec-Driven Development, Agentic Coding, AI Pair Programming, HITL, TDD+AI.
|
||||
|
||||
**Goal:** Establish Henry's expertise in AI-assisted development. Second article for Dev.to account warmup.
|
||||
|
||||
**Angle:** Survey + practitioner perspective (via interview with Oleg)
|
||||
|
||||
---
|
||||
|
||||
# Brief
|
||||
|
||||
*pending @strategist*
|
||||
|
||||
---
|
||||
|
||||
# Assets Index
|
||||
|
||||
All working files for this article:
|
||||
|
||||
| File | Purpose |
|
||||
|------|---------|
|
||||
| [outline.md](assets/beyond-vibe-coding/outline.md) | Article structure (pending) |
|
||||
| [text.md](assets/beyond-vibe-coding/text.md) | Article draft (pending) |
|
||||
| [seo-metadata.md](assets/beyond-vibe-coding/seo-metadata.md) | SEO title, description, keywords |
|
||||
| [log-chat.md](assets/beyond-vibe-coding/log-chat.md) | Activity log and agent comments |
|
||||
| [research-index.md](assets/beyond-vibe-coding/research-index.md) | Methodology clusters, verified sources, interview questions |
|
||||
| [interview.md](assets/beyond-vibe-coding/interview.md) | Oleg's answers — source for Henry's voice |
|
||||
|
||||
## External Research
|
||||
|
||||
| File | Purpose |
|
||||
|------|---------|
|
||||
| [perplexity-chats/AI-Assisted Development_...](research/perplexity-chats/) | Original Perplexity research on terminology |
|
||||
|
||||
---
|
||||
|
||||
# Activity Log
|
||||
|
||||
See [log-chat.md](assets/beyond-vibe-coding/log-chat.md)
|
||||
|
|
@ -0,0 +1,20 @@
|
|||
# Activity Log
|
||||
|
||||
## 2025-01-22 @strategist
|
||||
|
||||
**Action:** Initial setup, research intake
|
||||
|
||||
**Changes:**
|
||||
- Created article card in `0-inbox/`
|
||||
- Created assets folder structure
|
||||
- Copied Perplexity research to `perplexity-terminology-research.md`
|
||||
- Created `research-index.md` for methodology clustering
|
||||
|
||||
**Notes:**
|
||||
- Goal: warm-up article for Henry's Dev.to account
|
||||
- Approach: survey + personal opinion through interview with Oleg
|
||||
- Need to verify all source links before using
|
||||
|
||||
**For next agent:** @strategist continues with research clustering and interview
|
||||
|
||||
---
|
||||
|
|
@ -0,0 +1,3 @@
|
|||
# Outline
|
||||
|
||||
*pending — will be created after research clustering and interview*
|
||||
|
|
@ -0,0 +1,423 @@
|
|||
# AI-Assisted Development: Кластеризованная терминология и подходы
|
||||
|
||||
**Source:** Perplexity Research, January 2026
|
||||
**Original file:** `/research/perplexity-chats/AI-Assisted Development_ Кластеризованная терминол.md`
|
||||
|
||||
---
|
||||
|
||||
## Domain 1: Experimental & Low-Quality Approaches
|
||||
|
||||
### Vibe Coding
|
||||
|
||||
**Authority Rank: 1** | **Perception: Negative**
|
||||
|
||||
**Sources:**
|
||||
1. **Andrej Karpathy** (OpenAI co-founder, Tesla AI Director) — Wikipedia, Feb 2025
|
||||
2. **Collins English Dictionary** — Word of the Year 2025
|
||||
3. **SonarSource** (Code Quality Platform) — code quality analysis
|
||||
|
||||
**Description:**
|
||||
Term coined by Andrej Karpathy in February 2025, quickly became cultural phenomenon — Collins English Dictionary named it Word of Year 2025. Approach where developer describes task in natural language, AI generates code, but key distinction: **developer does NOT review code**, only looks at execution results.
|
||||
|
||||
Simon Willison quote: "If LLM wrote every line of your code, but you reviewed, tested and understood it all — that's not vibe coding, that's using LLM as typing assistant". Key characteristic: accepting AI-generated code without understanding it.
|
||||
|
||||
Critics point to lack of accountability, maintainability problems, increased security vulnerability risk. May 2025: Swedish app Lovable (using vibe coding) had security vulnerabilities in 170 of 1,645 created web apps. Fast Company September 2025 reported "vibe coding hangover" — senior engineers cite "development hell" working with such code.
|
||||
|
||||
Suitable for "throwaway weekend projects" as Karpathy originally intended, but risky for production systems.
|
||||
|
||||
**Links:**
|
||||
- https://en.wikipedia.org/wiki/Vibe_coding
|
||||
- https://www.sonarsource.com/resources/library/vibe-coding/
|
||||
|
||||
---
|
||||
|
||||
## Domain 2: Enterprise & Production-Grade Methodologies
|
||||
|
||||
### AI-Driven Development Life Cycle (AI-DLC)
|
||||
|
||||
**Authority Rank: 1** | **Perception: Positive - Enterprise**
|
||||
|
||||
**Sources:**
|
||||
1. **AWS** — Raja SP, Principal Solutions Architect, July 2025
|
||||
2. **Amazon Q Developer & Kiro** — official AWS platform
|
||||
|
||||
**Description:**
|
||||
Presented by AWS July 2025 as transformative enterprise methodology. Raja SP created AI-DLC with team after working with 100+ large customers.
|
||||
|
||||
AI as **central collaborator** throughout SDLC with two dimensions:
|
||||
1. **AI-Powered Execution with Human Oversight** — AI creates detailed work plans, actively requests clarifications, defers critical decisions to humans
|
||||
2. **Dynamic Team Collaboration** — while AI handles routine tasks, teams unite for real-time problem solving
|
||||
|
||||
**Three phases:** Inception (Mob Elaboration), Construction (Mob Construction), Operations
|
||||
|
||||
**Terminology:** "sprints" → "bolts" (hours/days instead of weeks); Epics → Units of Work
|
||||
|
||||
**Link:** https://aws.amazon.com/blogs/devops/ai-driven-development-life-cycle/
|
||||
|
||||
### Spec-Driven Development (SDD)
|
||||
|
||||
**Authority Rank: 2** | **Perception: Positive - Systematic**
|
||||
|
||||
**Sources:**
|
||||
1. **GitHub Engineering** — Den Delimarsky, Spec Kit toolkit, September 2025
|
||||
2. **ThoughtWorks Technology Radar** — November 2025
|
||||
3. **Red Hat Developers** — October 2025
|
||||
|
||||
**Description:**
|
||||
Emerged 2025 as direct response to "vibe coding" problems. ThoughtWorks included in Technology Radar. GitHub open-sourced Spec Kit September 2025, supports Claude Code, GitHub Copilot, Gemini CLI.
|
||||
|
||||
**Key principle:** specification becomes source of truth, not code.
|
||||
|
||||
**Spec Kit Workflow:**
|
||||
1. **Constitution** — immutable high-level principles (rules file)
|
||||
2. **/specify** — create specification from high-level prompt
|
||||
3. **/plan** — technical planning based on specification
|
||||
4. **/tasks** — break down into manageable phased parts
|
||||
|
||||
**Three interpretations:** Spec-first, Spec-anchored, Spec-as-source
|
||||
|
||||
**Tools:** Amazon Kiro, GitHub Spec Kit, Tessl Framework
|
||||
|
||||
**Links:**
|
||||
- https://github.blog/ai-and-ml/generative-ai/spec-driven-development-with-ai-get-started-with-a-new-open-source-toolkit/
|
||||
- https://www.thoughtworks.com/radar/techniques/spec-driven-development
|
||||
- https://martinfowler.com/articles/exploring-gen-ai/sdd-3-tools.html
|
||||
- https://developer.microsoft.com/blog/spec-driven-development-spec-kit
|
||||
|
||||
### Architecture-First AI Development
|
||||
|
||||
**Authority Rank: 3** | **Perception: Positive - Professional/Mature**
|
||||
|
||||
**Sources:**
|
||||
1. **WaveMaker** — Vikram Srivats (CCO), Prashant Reddy (Head of AI Product Engineering), January 2026
|
||||
2. **ITBrief Industry Analysis** — January 2026
|
||||
|
||||
**Description:**
|
||||
2026 industry shift. Quote Vikram Srivats: "Second coming of AI coding tools must be all about **Architectural Intelligence** — just Artificial Intelligence no longer fits."
|
||||
|
||||
Shift from "vibe coding" experiments to governance, architecture alignment, long-term maintainability.
|
||||
|
||||
**Key characteristics:**
|
||||
- System design before implementation
|
||||
- AI agents with clear roles: Architect, Builder, Guardian
|
||||
- Coding architectural rules, enforcement review processes
|
||||
- Working from formal specifications
|
||||
- Respect for internal organizational standards
|
||||
|
||||
**Links:**
|
||||
- https://itbrief.co.uk/story/ai-coding-tools-face-2026-reset-towards-architecture
|
||||
- https://itbrief.news/story/ai-coding-tools-face-2026-reset-towards-architecture
|
||||
|
||||
---
|
||||
|
||||
## Domain 3: Quality & Validation-Focused Approaches
|
||||
|
||||
### Test-Driven Development with AI (TDD-AI)
|
||||
|
||||
**Authority Rank: 1** | **Perception: Positive - Quality-Focused**
|
||||
|
||||
**Sources:**
|
||||
1. **Galileo AI Research** — August 2025
|
||||
2. **Builder.io Engineering** — August 2025
|
||||
|
||||
**Description:**
|
||||
Traditional TDD adapted for AI systems. Tests written first → AI generates code to pass tests → verify → refactor.
|
||||
|
||||
Statistical testing for non-deterministic AI outputs — critical distinction from traditional TDD.
|
||||
|
||||
**Links:**
|
||||
- https://galileo.ai/blog/tdd-ai-system-architecture
|
||||
- https://galileo.ai/blog/test-driven-development-ai-systems
|
||||
- https://www.builder.io/blog/test-driven-development-ai
|
||||
|
||||
### Human-in-the-Loop (HITL) AI Development
|
||||
|
||||
**Authority Rank: 2** | **Perception: Positive - Responsible**
|
||||
|
||||
**Sources:**
|
||||
1. **Google Cloud Documentation** — 2026
|
||||
2. **Encord Research** — December 2024
|
||||
3. **Atlassian Engineering** — HULA framework, September 2025
|
||||
|
||||
**Description:**
|
||||
Humans actively involved in AI system lifecycle. Continuous feedback and validation loops. Hybrid approach: human judgment + AI execution.
|
||||
|
||||
**HULA (Human-in-the-Loop AI)** — Atlassian framework for software development agents.
|
||||
|
||||
**Links:**
|
||||
- https://cloud.google.com/discover/human-in-the-loop
|
||||
- https://encord.com/blog/human-in-the-loop-ai/
|
||||
- https://www.atlassian.com/blog/atlassian-engineering/hula-blog-autodev-paper-human-in-the-loop-software-development-agents
|
||||
|
||||
### Quality-First AI Coding
|
||||
|
||||
**Authority Rank: 3** | **Perception: Positive - Professional**
|
||||
|
||||
**Sources:**
|
||||
1. **Qodo.ai** (formerly CodiumAI) — December 2025
|
||||
|
||||
**Description:**
|
||||
Code integrity at core. Qodo.ai — platform with agentic AI code generation and comprehensive testing.
|
||||
|
||||
Production-ready focus: automatic test generation for every code change. Direct contrast to "vibe coding" — quality non-negotiable.
|
||||
|
||||
**Link:** https://www.qodo.ai/ai-code-review-platform/
|
||||
|
||||
### Deterministic AI Development
|
||||
|
||||
**Authority Rank: 4** | **Perception: Positive - Enterprise/Compliance**
|
||||
|
||||
**Source:** Augment Code Research — August 2025
|
||||
|
||||
**Description:**
|
||||
Identical outputs for identical inputs. Rule-based architectures for predictability. Best for: security scanning, compliance checks, refactoring tasks.
|
||||
|
||||
Hybrid approach: probabilistic reasoning + deterministic execution.
|
||||
|
||||
**Link:** https://www.augmentcode.com/guides/deterministic-ai-for-predictable-coding
|
||||
|
||||
---
|
||||
|
||||
## Domain 4: Collaborative Development Patterns
|
||||
|
||||
### AI Pair Programming
|
||||
|
||||
**Authority Rank: 1** | **Perception: Positive - Collaborative**
|
||||
|
||||
**Sources:**
|
||||
1. **GitHub Copilot (Microsoft)** — January 2026
|
||||
2. **Qodo.ai Documentation** — March 2025
|
||||
3. **GeeksforGeeks** — July 2025
|
||||
|
||||
**Description:**
|
||||
AI as "pair programmer" or coding partner. Based on traditional pair programming: driver (human/AI) and navigator (human/AI) roles.
|
||||
|
||||
Real-time collaboration and feedback. Tools: GitHub Copilot, Cursor, Windsurf.
|
||||
|
||||
**Links:**
|
||||
- https://code.visualstudio.com/docs/copilot/overview
|
||||
- https://www.qodo.ai/glossary/pair-programming/
|
||||
- https://www.geeksforgeeks.org/artificial-intelligence/what-is-ai-pair-programming/
|
||||
- https://graphite.com/guides/ai-pair-programming-best-practices
|
||||
|
||||
### Mobbing with AI / Mob Programming with AI
|
||||
|
||||
**Authority Rank: 2** | **Perception: Positive - Team-Focused**
|
||||
|
||||
**Sources:**
|
||||
1. **Atlassian Engineering Blog** — December 2025
|
||||
2. **Aaron Griffith** — January 2025 (YouTube)
|
||||
|
||||
**Description:**
|
||||
Entire team works together, AI as driver. AI generates code/tests in front of team. Team navigates, reviews, refines in real-time.
|
||||
|
||||
Best for: complex problems, knowledge transfer, quality assurance.
|
||||
|
||||
**Links:**
|
||||
- https://www.atlassian.com/blog/atlassian-engineering/mobbing-with-ai
|
||||
- https://www.youtube.com/watch?v=BsFPbYX4WXQ
|
||||
|
||||
### Agentic Coding / Agentic Programming
|
||||
|
||||
**Authority Rank: 3** | **Perception: Positive - Advanced**
|
||||
|
||||
**Sources:**
|
||||
1. **arXiv Research Paper** — August 2025
|
||||
2. **AI Accelerator Institute** — February 2025
|
||||
3. **Apiiro Security Platform** — September 2025
|
||||
|
||||
**Description:**
|
||||
LLM-based agents autonomously plan, execute, improve development tasks. Beyond code completion: generates programs, diagnoses bugs, writes tests, refactors.
|
||||
|
||||
Key properties: **autonomy, interactive, iterative refinement, goal-oriented**.
|
||||
|
||||
Agent behaviors: planning, memory management, tool integration, execution monitoring.
|
||||
|
||||
**Links:**
|
||||
- https://arxiv.org/html/2508.11126v1
|
||||
- https://www.aiacceleratorinstitute.com/agentic-code-generation-the-future-of-software-development/
|
||||
- https://apiiro.com/glossary/agentic-coding/
|
||||
|
||||
---
|
||||
|
||||
## Domain 5: Workflow & Process Integration
|
||||
|
||||
### Prompt-Driven Development (PDD)
|
||||
|
||||
**Authority Rank: 1** | **Perception: Neutral to Positive**
|
||||
|
||||
**Sources:**
|
||||
1. **Capgemini Software Engineering** — May 2025
|
||||
2. **Hexaware Technologies** — August 2025
|
||||
|
||||
**Description:**
|
||||
Developer breaks requirements into series of prompts. LLM generates code for each prompt. **Critical:** developer MUST review LLM-generated code.
|
||||
|
||||
Critical distinction from vibe coding: code review mandatory.
|
||||
|
||||
**Links:**
|
||||
- https://capgemini.github.io/ai/prompt-driven-development/
|
||||
- https://hexaware.com/blogs/prompt-driven-development-coding-in-conversation/
|
||||
|
||||
### AI-Augmented Development
|
||||
|
||||
**Authority Rank: 2** | **Perception: Positive - Practical**
|
||||
|
||||
**Sources:**
|
||||
1. **GitLab Official Documentation** — December 2023
|
||||
2. **Virtusa** — January 2024
|
||||
|
||||
**Description:**
|
||||
AI tools accelerate SDLC across all phases. Focus: code generation, bug detection, automated testing, smart documentation.
|
||||
|
||||
Key principle: humans handle strategy, AI handles execution.
|
||||
|
||||
**Links:**
|
||||
- https://about.gitlab.com/topics/agentic-ai/ai-augmented-software-development/
|
||||
- https://www.virtusa.com/digital-themes/ai-augmented-development
|
||||
|
||||
### Copilot-Driven Development
|
||||
|
||||
**Authority Rank: 3** | **Perception: Positive - Practical**
|
||||
|
||||
**Sources:**
|
||||
1. **GitHub/Microsoft Official** — January 2026
|
||||
2. **Emergn** — September 2025
|
||||
|
||||
**Description:**
|
||||
Specifically using GitHub Copilot or similar tools as development partner (not just assistant).
|
||||
|
||||
Context-aware, learns coding style. Enables conceptual focus instead of mechanical typing.
|
||||
|
||||
**Links:**
|
||||
- https://code.visualstudio.com/docs/copilot/overview
|
||||
- https://www.emergn.com/insights/how-ai-tools-impact-the-way-we-develop-software-our-github-copilot-journey/
|
||||
|
||||
### Conversational Coding
|
||||
|
||||
**Authority Rank: 4** | **Perception: Neutral to Positive**
|
||||
|
||||
**Sources:**
|
||||
1. **Google Cloud Platform** — January 2026
|
||||
2. **arXiv Research** — March 2025
|
||||
|
||||
**Description:**
|
||||
Natural language interaction with AI for development. Iterative, dialogue-based approach. Context retention across sessions.
|
||||
|
||||
**Links:**
|
||||
- https://cloud.google.com/conversational-ai
|
||||
- https://arxiv.org/abs/2503.16508
|
||||
|
||||
---
|
||||
|
||||
## Domain 6: Code Review & Maintenance
|
||||
|
||||
### AI Code Review
|
||||
|
||||
**Authority Rank: 1** | **Perception: Neutral to Positive**
|
||||
|
||||
**Sources:**
|
||||
1. **LinearB** — March 2024
|
||||
2. **Swimm.io** — November 2025
|
||||
3. **CodeAnt.ai** — May 2025
|
||||
|
||||
**Description:**
|
||||
Automated code examination using ML/LLM. Static and dynamic analysis. Identifies bugs, security issues, performance problems, code smells.
|
||||
|
||||
Tools: Qodo, CodeRabbit, SonarQube AI features.
|
||||
|
||||
**Links:**
|
||||
- https://linearb.io/blog/ai-code-review
|
||||
- https://swimm.io/learn/ai-tools-for-developers/ai-code-review-how-it-works-and-3-tools-you-should-know
|
||||
- https://www.codeant.ai/blogs/ai-vs-traditional-code-review
|
||||
|
||||
---
|
||||
|
||||
## Domain 7: Specialized & Emerging Approaches
|
||||
|
||||
### Ensemble Programming/Prompting with AI
|
||||
|
||||
**Authority Rank: 1** | **Perception: Positive - Advanced**
|
||||
|
||||
**Sources:**
|
||||
1. **Kinde.com** — November 2024
|
||||
2. **Ultralytics ML Research** — December 2025
|
||||
3. **arXiv** — June 2025
|
||||
|
||||
**Description:**
|
||||
Multiple AI models/prompts combined for better results. Aggregation methods: voting, averaging, weighted scoring.
|
||||
|
||||
**Links:**
|
||||
- https://kinde.com/learn/ai-for-software-engineering/prompting/ensemble-prompting-that-actually-moves-the-needle/
|
||||
- https://www.ultralytics.com/blog/exploring-ensemble-learning-and-its-role-in-ai-and-ml
|
||||
|
||||
### Prompt Engineering for Development
|
||||
|
||||
**Authority Rank: 2** | **Perception: Neutral to Positive**
|
||||
|
||||
**Sources:**
|
||||
1. **Google Cloud** — January 2026
|
||||
2. **OpenAI** — April 2025
|
||||
3. **GitHub** — May 2024
|
||||
|
||||
**Description:**
|
||||
Crafting effective prompts for AI models. Critical skill for AI-assisted development.
|
||||
|
||||
Techniques: few-shot learning, chain-of-thought, role prompting.
|
||||
|
||||
**Links:**
|
||||
- https://cloud.google.com/discover/what-is-prompt-engineering
|
||||
- https://platform.openai.com/docs/guides/prompt-engineering
|
||||
- https://github.blog/ai-and-ml/generative-ai/prompt-engineering-guide-generative-ai-llms/
|
||||
|
||||
### Intentional AI Development
|
||||
|
||||
**Authority Rank: 3** | **Perception: Positive - Thoughtful**
|
||||
|
||||
**Sources:**
|
||||
1. **Tech.eu** — January 2026
|
||||
2. **ghuntley.com** — August 2025
|
||||
|
||||
**Description:**
|
||||
Purpose-driven AI design. Clear roles and boundaries for AI. Deliberate practice and learning approach.
|
||||
|
||||
**Links:**
|
||||
- https://tech.eu/2026/01/05/adopting-an-intentional-ai-strategy-in-2026/
|
||||
- https://ghuntley.com/play/
|
||||
|
||||
---
|
||||
|
||||
## Domain 8: General & Cross-Cutting Terms
|
||||
|
||||
### AI-Assisted Coding / AI-Assisted Development
|
||||
|
||||
**Authority Rank: 1** | **Perception: Neutral to Positive**
|
||||
|
||||
**Sources:**
|
||||
1. **Wikipedia** — July 2025
|
||||
2. **GitLab** — 2025
|
||||
|
||||
**Description:**
|
||||
Broad umbrella term for AI enhancing software development tasks. Includes code completion, documentation generation, testing, debugging assistance.
|
||||
|
||||
Developer remains in control, reviews all suggestions. Most common adoption pattern globally.
|
||||
|
||||
**Links:**
|
||||
- https://en.wikipedia.org/wiki/AI-assisted_software_development
|
||||
- https://about.gitlab.com/topics/devops/ai-code-generation-guide/
|
||||
|
||||
---
|
||||
|
||||
## Key Takeaways
|
||||
|
||||
**Domain 1** (Experimental): Only Vibe Coding — only term with explicitly negative connotation, backed by high-authority sources (OpenAI founder, Collins Dictionary).
|
||||
|
||||
**Domain 2** (Enterprise): Most authoritative domain with AWS, GitHub Engineering, ThoughtWorks as sources. Focus on production-grade, governance, architecture.
|
||||
|
||||
**Domain 3** (Quality): Research-heavy domain (Galileo AI, Google Cloud, Atlassian) with emphasis on responsible development.
|
||||
|
||||
**Domain 4** (Collaborative): Practical patterns, backed by major platforms (Microsoft/GitHub, Atlassian) and research (arXiv).
|
||||
|
||||
**Domains 5-7**: Workflow integration, code review, specialized techniques — more narrow but important practices.
|
||||
|
||||
**Domain 8**: General term serving as baseline for all other approaches.
|
||||
|
|
@ -0,0 +1,79 @@
|
|||
# Research Index
|
||||
|
||||
Working file for methodology clustering and validation.
|
||||
|
||||
## Source
|
||||
- `perplexity-terminology-research.md` — original Perplexity research (Jan 2025)
|
||||
|
||||
---
|
||||
|
||||
## Methodology Clusters
|
||||
|
||||
### Tier 1: Must Include (High Authority + High Relevance)
|
||||
|
||||
| Term | Source Authority | Link Status | Include? |
|
||||
|------|------------------|-------------|----------|
|
||||
| Vibe Coding | Wikipedia, Collins Dictionary, Andrej Karpathy | ⏳ verify | |
|
||||
| Spec-Driven Development | GitHub, ThoughtWorks, Martin Fowler | ⏳ verify | |
|
||||
| AI-Driven Development Life Cycle | AWS | ⏳ verify | |
|
||||
| Agentic Coding | arXiv, AI Accelerator Institute | ⏳ verify | |
|
||||
| AI Pair Programming | GitHub/Microsoft, GeeksforGeeks | ⏳ verify | |
|
||||
|
||||
### Tier 2: Consider (Good Authority)
|
||||
|
||||
| Term | Source Authority | Link Status | Include? |
|
||||
|------|------------------|-------------|----------|
|
||||
| Architecture-First AI Development | WaveMaker, ITBrief | ⏳ verify | |
|
||||
| Test-Driven Development with AI | Galileo AI, Builder.io | ⏳ verify | |
|
||||
| Human-in-the-Loop (HITL) | Google Cloud, Atlassian | ⏳ verify | |
|
||||
| Prompt-Driven Development | Capgemini, Hexaware | ⏳ verify | |
|
||||
| Quality-First AI Coding | Qodo.ai | ⏳ verify | |
|
||||
|
||||
### Tier 3: Maybe (Lower Authority / Niche)
|
||||
|
||||
| Term | Source Authority | Link Status | Include? |
|
||||
|------|------------------|-------------|----------|
|
||||
| Mobbing with AI | Atlassian | ⏳ verify | |
|
||||
| Copilot-Driven Development | Microsoft/GitHub | ⏳ verify | |
|
||||
| Conversational Coding | Google Cloud, arXiv | ⏳ verify | |
|
||||
| Deterministic AI Development | Augment Code | ⏳ verify | |
|
||||
| Ensemble Programming | Kinde.com, Ultralytics | ⏳ verify | |
|
||||
|
||||
### Tier 4: Skip (Generic / Low Value)
|
||||
|
||||
- AI-Assisted Coding — too generic, umbrella term
|
||||
- AI-Augmented Development — too generic
|
||||
- Prompt Engineering — separate skill, not methodology
|
||||
- AI Code Review — tool category, not methodology
|
||||
|
||||
---
|
||||
|
||||
## Interview Questions Bank
|
||||
|
||||
*Questions to ask Oleg about each methodology*
|
||||
|
||||
1. Vibe Coding — "Ты когда-нибудь так работал? Результат?"
|
||||
2. Spec-Driven — "Пробовал писать spec перед кодом с AI?"
|
||||
3. Agentic — "Как используешь Claude Code в agentic режиме?"
|
||||
4. TDD-AI — "Тесты сначала с AI — работает?"
|
||||
5. HITL — "Как часто AI делает что-то без твоего OK?"
|
||||
|
||||
---
|
||||
|
||||
## Henry's Opinions (from interview)
|
||||
|
||||
*Will be filled during interview*
|
||||
|
||||
---
|
||||
|
||||
## Link Verification Log
|
||||
|
||||
*Track which links checked and status*
|
||||
|
||||
| Link | Status | Notes |
|
||||
|------|--------|-------|
|
||||
| | | |
|
||||
|
||||
---
|
||||
|
||||
*Updated: 2025-01-22*
|
||||
|
|
@ -0,0 +1,16 @@
|
|||
# SEO Metadata
|
||||
|
||||
*pending — will be created after keyword research (step 5)*
|
||||
|
||||
## Title
|
||||
[TBD]
|
||||
|
||||
## Meta Description
|
||||
[TBD]
|
||||
|
||||
## Target Keywords
|
||||
- Primary: [TBD]
|
||||
- Secondary: [TBD]
|
||||
|
||||
## URL Slug
|
||||
ai-coding-methodologies-beyond-vibe-coding
|
||||
|
|
@ -0,0 +1,3 @@
|
|||
# Article Text
|
||||
|
||||
*pending — will be written after outline approval*
|
||||
|
|
@ -0,0 +1,74 @@
|
|||
# Interview Notes
|
||||
|
||||
Записи ответов Олега на вопросы по методологиям AI-разработки.
|
||||
Используется для создания голоса Henry в статье.
|
||||
|
||||
---
|
||||
|
||||
## Vibe Coding
|
||||
|
||||
**Вопрос:** Ты когда-нибудь работал в режиме vibe coding? Что получилось? Когда это уместно, когда нет?
|
||||
|
||||
**Ответ:**
|
||||
|
||||
Термину меньше года, а он уже стал практически синонимом AI-разработки. Наверное потому что удобно звучит — "vibe coding", "я навайбкодил". Но в этом и проблема: vibe coding сохраняет негативные коннотации (недостаточно профессионально, ненадёжно, абы-как), и при этом обобщает весь AI-кодинг, перенося этот негатив на всё поле. Поэтому и пишу эту статью — давайте разберём где vibe coding, а где другие подходы.
|
||||
|
||||
**Когда использую:** Dev tools не попадающие в прод, прототипы, эксперименты, side projects. Надо признать — во многих случаях работает неплохо.
|
||||
|
||||
**Но не чистый vibe:** Обычно всё равно проверяю изменения — хотя бы быстро просматриваю git diff перед коммитом. Если задача большая — прошу агента коммитить небольшими порциями, потом просматриваю.
|
||||
|
||||
**Хорошие практики даже при vibe coding:**
|
||||
1. Покрывать код тестами и проверять что проходят (+ typecheck, lint, prettier)
|
||||
2. Просить другого AI агента сделать ревью проделанной работы
|
||||
3. Человеческое внимание для важных вещей — если что-то критично, лучше убедиться самостоятельно
|
||||
|
||||
Не говорю что нужно всегда досконально проверять если код работает, но минимальный контроль — да.
|
||||
|
||||
---
|
||||
|
||||
## Spec-Driven Development
|
||||
|
||||
**Вопрос:** Пробовал GitHub Spec Kit или писать spec перед кодом? Это будущее или overkill?
|
||||
|
||||
**Ответ:**
|
||||
[pending]
|
||||
|
||||
---
|
||||
|
||||
## Agentic Coding
|
||||
|
||||
**Вопрос:** Как используешь Claude Code? Даёшь автономию или контролируешь каждый шаг? Что доверяешь агенту, что нет?
|
||||
|
||||
**Ответ:**
|
||||
[pending]
|
||||
|
||||
---
|
||||
|
||||
## AI Pair Programming
|
||||
|
||||
**Вопрос:** Copilot/Cursor — как pair programmer или просто autocomplete? Реально ли это "парное программирование"?
|
||||
|
||||
**Ответ:**
|
||||
[pending]
|
||||
|
||||
---
|
||||
|
||||
## Human-in-the-Loop
|
||||
|
||||
**Вопрос:** Как часто AI делает что-то без твоего одобрения? Где граница доверия?
|
||||
|
||||
**Ответ:**
|
||||
[pending]
|
||||
|
||||
---
|
||||
|
||||
## TDD + AI
|
||||
|
||||
**Вопрос:** Пишешь тесты первыми при работе с AI? Работает ли классический TDD с генеративным AI?
|
||||
|
||||
**Ответ:**
|
||||
[pending]
|
||||
|
||||
---
|
||||
|
||||
*Created: 2026-01-22*
|
||||
|
|
@ -0,0 +1,20 @@
|
|||
# Activity Log
|
||||
|
||||
## 2026-01-22 @strategist
|
||||
|
||||
**Action:** Initial setup
|
||||
|
||||
**Changes:**
|
||||
- Created article card in `0-inbox/beyond-vibe-coding.md`
|
||||
- Created assets folder structure
|
||||
- Copied Perplexity research
|
||||
- Created research-index.md for clustering
|
||||
|
||||
**Notes:**
|
||||
- Goal: Henry's 2nd Dev.to article for account warmup
|
||||
- Approach: methodology survey + practitioner opinion via interview
|
||||
- Interview planned to capture authentic perspective
|
||||
|
||||
**Next:** Verify sources, cluster methodologies, conduct interview
|
||||
|
||||
---
|
||||
|
|
@ -0,0 +1,3 @@
|
|||
# Outline
|
||||
|
||||
*pending — will be created after research clustering and interview*
|
||||
|
|
@ -0,0 +1,166 @@
|
|||
# Research Index
|
||||
|
||||
Working file for methodology clustering and source verification.
|
||||
|
||||
## Source
|
||||
- `perplexity-terminology-research.md` — original Perplexity research (Jan 2025)
|
||||
|
||||
---
|
||||
|
||||
## Methodology Selection
|
||||
|
||||
Based on research, selecting approaches for article coverage.
|
||||
|
||||
### Tier 1: MUST Include (High Authority + Reader Interest)
|
||||
|
||||
| Term | Why Include | Source Authority | Status |
|
||||
|------|-------------|------------------|--------|
|
||||
| **Vibe Coding** | Hook — Collins Word of Year 2025, everyone knows it | Wikipedia, Collins, Karpathy | ✅ verified |
|
||||
| **Spec-Driven Development** | Direct contrast to vibe coding, GitHub Spec Kit | GitHub, ThoughtWorks, Martin Fowler | ✅ verified |
|
||||
| **Agentic Coding** | Hot topic, Claude Code users care about this | arXiv surveys (Aug 2025, Dec 2025) | ✅ verified |
|
||||
|
||||
### Tier 2: SHOULD Include (Good Authority, Practical)
|
||||
|
||||
| Term | Why Include | Source Authority | Status |
|
||||
|------|-------------|------------------|--------|
|
||||
| **AI Pair Programming** | Relatable, Copilot users know this | GitHub official, Microsoft | ✅ verified |
|
||||
| **Human-in-the-Loop (HITL)** | Important concept, enterprise backing | Atlassian HULA, Google Cloud | ✅ verified |
|
||||
| **TDD + AI** | Quality angle, contrasts vibe coding | Qodo.ai, Builder.io, GitHub Blog | ✅ verified |
|
||||
|
||||
### Tier 3: MENTION Briefly (Context/Completeness)
|
||||
|
||||
| Term | Why | Status |
|
||||
|------|-----|--------|
|
||||
| AI-DLC (AWS) | Enterprise angle, shows maturity | ⏳ verify if needed |
|
||||
| Architecture-First | 2026 trend prediction | ⏳ verify if needed |
|
||||
| Prompt-Driven Development | Clarify difference from vibe coding | ⏳ verify if needed |
|
||||
|
||||
### Tier 4: SKIP (Generic/Low Value for This Article)
|
||||
|
||||
- AI-Assisted Coding — umbrella term, too generic
|
||||
- AI-Augmented Development — same, no distinct methodology
|
||||
- Copilot-Driven Development — subset of AI Pair Programming
|
||||
- Conversational Coding — too vague
|
||||
- Ensemble Programming — too niche
|
||||
- Deterministic AI Development — enterprise-specific, niche
|
||||
|
||||
---
|
||||
|
||||
## Article Angle
|
||||
|
||||
**Narrative arc:**
|
||||
1. Vibe coding = entry point (hot topic, word of year)
|
||||
2. Problem: works for prototypes, fails for production
|
||||
3. Survey: what professionals actually use
|
||||
4. Each methodology = one section with:
|
||||
- What it is (1-2 sentences)
|
||||
- When to use
|
||||
- Henry's take (from interview)
|
||||
5. Conclusion: spectrum from vibe → spec-driven, choose based on stakes
|
||||
|
||||
---
|
||||
|
||||
## Verified Sources
|
||||
|
||||
### Vibe Coding
|
||||
| Source | URL | Notes |
|
||||
|--------|-----|-------|
|
||||
| Wikipedia | en.wikipedia.org/wiki/Vibe_coding | Comprehensive, cites Karpathy |
|
||||
| Collins Dictionary | collinsdictionary.com | Word of Year 2025 |
|
||||
| Andrej Karpathy | X/Twitter Feb 2025 | Original coiner |
|
||||
|
||||
### Spec-Driven Development
|
||||
| Source | URL | Notes |
|
||||
|--------|-----|-------|
|
||||
| GitHub Spec Kit | github.com/github/spec-kit | Official repo |
|
||||
| ThoughtWorks Radar | thoughtworks.com/radar | SDD in techniques |
|
||||
| Martin Fowler | martinfowler.com | SDD tools analysis |
|
||||
|
||||
### Agentic Coding
|
||||
| Source | URL | Notes |
|
||||
|--------|-----|-------|
|
||||
| arXiv 2508.11126 | arxiv.org/abs/2508.11126 | "AI Agentic Programming" survey (Aug 2025) |
|
||||
| arXiv 2512.07921 | arxiv.org/abs/2512.07921 | "DeepCode: Open Agentic Coding" (Dec 2025) |
|
||||
| arXiv 2512.14012 | arxiv.org/html/2512.14012 | "Professional Developers Don't Vibe, They Control" |
|
||||
|
||||
### AI Pair Programming
|
||||
| Source | URL | Notes |
|
||||
|--------|-----|-------|
|
||||
| GitHub Features | github.com/features/copilot | "Your AI pair programmer" |
|
||||
| GitHub Blog | github.blog (June 2025) | "From pair to peer programmer" |
|
||||
| Microsoft Visual Studio | visualstudio.microsoft.com/github-copilot | Official docs |
|
||||
| Microsoft Learn | learn.microsoft.com/training/paths/copilot | Training courses |
|
||||
|
||||
### Human-in-the-Loop
|
||||
| Source | URL | Notes |
|
||||
|--------|-----|-------|
|
||||
| Atlassian HULA | atlassian.com/blog/atlassian-engineering/hula-blog | ICSE 2025 paper |
|
||||
| arXiv 2411.12924 | arxiv.org/abs/2411.12924 | HULA framework paper |
|
||||
| Google Cloud | cloud.google.com/discover/human-in-the-loop | Official definition |
|
||||
| Google Developers Blog | developers.googleblog.com | A2A protocol with HITL |
|
||||
|
||||
### TDD + AI
|
||||
| Source | URL | Notes |
|
||||
|--------|-----|-------|
|
||||
| Qodo.ai | qodo.ai/blog/ai-code-assistants-test-driven-development | AWS SA author |
|
||||
| Builder.io | builder.io/blog/test-driven-development-ai | Practical guide |
|
||||
| GitHub Blog | github.blog (May 2025) | "TDD with GitHub Copilot" |
|
||||
| Substack | ecosystem4engineering.substack.com | "TDD in Age of AI" |
|
||||
|
||||
---
|
||||
|
||||
## Interview Questions
|
||||
|
||||
Questions for Oleg to capture Henry's authentic voice:
|
||||
|
||||
### Vibe Coding
|
||||
- Ты когда-нибудь работал в режиме vibe coding? Что получилось?
|
||||
- Когда это уместно, когда нет?
|
||||
|
||||
### Spec-Driven Development
|
||||
- Пробовал GitHub Spec Kit или писать spec перед кодом?
|
||||
- Твоё мнение — это будущее или overkill?
|
||||
|
||||
### Agentic Coding
|
||||
- Как используешь Claude Code? Даёшь автономию или контролируешь каждый шаг?
|
||||
- Что доверяешь агенту, что нет?
|
||||
|
||||
### AI Pair Programming
|
||||
- Copilot/Cursor — как pair programmer или просто autocomplete?
|
||||
- Реально ли это "парное программирование"?
|
||||
|
||||
### Human-in-the-Loop
|
||||
- Как часто AI делает что-то без твоего одобрения?
|
||||
- Где граница доверия?
|
||||
|
||||
### TDD + AI
|
||||
- Пишешь тесты первыми при работе с AI?
|
||||
- Работает ли классический TDD с генеративным AI?
|
||||
|
||||
---
|
||||
|
||||
## Henry's Opinions (from interview)
|
||||
|
||||
*Will be filled during interview session*
|
||||
|
||||
### On Vibe Coding
|
||||
[pending]
|
||||
|
||||
### On Spec-Driven Development
|
||||
[pending]
|
||||
|
||||
### On Agentic Coding
|
||||
[pending]
|
||||
|
||||
### On AI Pair Programming
|
||||
[pending]
|
||||
|
||||
### On Human-in-the-Loop
|
||||
[pending]
|
||||
|
||||
### On TDD + AI
|
||||
[pending]
|
||||
|
||||
---
|
||||
|
||||
*Updated: 2026-01-22*
|
||||
|
|
@ -0,0 +1,16 @@
|
|||
# SEO Metadata
|
||||
|
||||
*pending — will be created after keyword research*
|
||||
|
||||
## Title
|
||||
[TBD]
|
||||
|
||||
## Meta Description
|
||||
[TBD]
|
||||
|
||||
## Target Keywords
|
||||
- Primary: [TBD]
|
||||
- Secondary: [TBD]
|
||||
|
||||
## URL Slug
|
||||
beyond-vibe-coding
|
||||
|
|
@ -0,0 +1,3 @@
|
|||
# Article Text
|
||||
|
||||
*pending — will be written after outline approval*
|
||||
Loading…
Reference in New Issue