From c5fad9f0924553f6f10dccb3763d2ebec04a7d7d Mon Sep 17 00:00:00 2001 From: Florian BRUNIAUX Date: Fri, 6 Feb 2026 16:09:02 +0100 Subject: [PATCH] docs: add Context Engineering (Thoughtworks) + corporate marketplaces footnotes - Add Context Engineering framework reference (Thoughtworks Tech Radar Vol 33) - Add emerging corporate AI marketplaces concept (Hugo 2026) - Document evaluation in docs/resource-evaluations/hugo-ai-impact-2026.md - Score: 2/5 (marginal) - minimal integration via footnotes only Co-Authored-By: Claude Sonnet 4.5 --- .../hugo-ai-impact-2026.md | 198 ++++++++++++++++++ guide/adoption-approaches.md | 4 + guide/methodologies.md | 4 + 3 files changed, 206 insertions(+) create mode 100644 docs/resource-evaluations/hugo-ai-impact-2026.md diff --git a/docs/resource-evaluations/hugo-ai-impact-2026.md b/docs/resource-evaluations/hugo-ai-impact-2026.md new file mode 100644 index 0000000..241c6cb --- /dev/null +++ b/docs/resource-evaluations/hugo-ai-impact-2026.md @@ -0,0 +1,198 @@ +# Resource Evaluation: AI's Impact on Software Engineering in 2026 + +**URL**: https://eventuallymaking.io/p/ai-s-impact-on-the-state-of-the-art-in-software-engineering-in-2026 +**Author**: Hugo (Software Engineer, 20+ years, Founder Malt/Writizzy) +**Published**: February 6, 2026 +**Type**: Opinion article based on 7 French company interviews +**Evaluated**: February 6, 2026 +**Evaluator**: Claude Code Guide Team + +--- + +## Summary + +Opinion piece on AI's impact on software engineering practices in 2026, based on interviews with 7 French tech companies (Doctolib, Malt, Alan, Google Cloud, Brevo, ManoMano, Ilek, Clever Cloud). + +**Key arguments**: +1. **Context Engineering** (Thoughtworks framework) — shift toward complete specifications with constraints before coding +2. **Spec/Plan/Act workflow standardization** — industry consensus on 3-phase approach +3. **Corporate AI governance** — organizational marketplaces to pool AI skills, agents, rules +4. **QA via CI/CD** — traditional practices (linting, testing, review) essential for AI-generated code validation +5. **HR disruption** — junior training, recruitment, career trajectories require restructuring + +**Stats cited**: +- Monthly costs: ~$20/dev (adoption), ~$200/dev (strong adoption), $200-1000+/dev (advanced multi-agent) +- 90%+ of engineers at Alan use AI-powered coding assistants daily +- Interviews: 7 companies (no detailed verbatims provided) + +--- + +## Evaluation Scores + +| Criterion | Score (1-5) | Notes | +|-----------|-------------|-------| +| **Relevance** | 2 | Marginal — concepts largely covered in guide, one legitimate gap (Context Engineering) | +| **Accuracy** | 3 | Moderate — terminology error ("Context Driven" vs "Context Engineering"), stats lack methodology | +| **Actionability** | 1 | Low — no templates, code, or concrete workflows | +| **Novelty** | 2 | Marginal — Spec/Plan/Act and QA/CI/CD already in guide, Context Engineering framework new | +| **Production-Ready** | 1 | Low — opinion piece, no implementation details | + +**Overall Score**: **2/5** (Marginal - Info secondaire) + +--- + +## Gap Analysis + +### What's NEW (not in guide) + +| Aspect | Hugo's Resource | Our Guide | Gap? | +|--------|----------------|-----------|------| +| **Context Engineering** (Thoughtworks) | ✓ Mentioned (but miscited as "Context Driven") | ✗ Absent | ✅ **Legitimate gap** | +| **Corporate AI marketplaces** | ✓ Concept described | ✗ Not covered | ⚠️ **Minor gap** (RH focus, not technical) | + +### What's ALREADY COVERED + +| Aspect | Hugo's Resource | Our Guide | +|--------|----------------|-----------| +| Spec/Plan/Act workflow | ✓ Described | ✅ `guide/workflows/spec-first.md`, `/plan` mode | +| QA via CI/CD | ✓ Mentioned | ✅ `guide/production-safety.md`, hooks | +| HR/Junior disruption | ✓ Opinion | ✅ `guide/learning-with-ai.md` (comprehensive) | +| Cost estimates | ✓ Ranges ($20-1000) | ✅ `guide/ai-ecosystem.md` (precise: $20-50) | + +--- + +## Fact-Check Results + +| Claim | Verified | Source | Correction | +|-------|----------|--------|------------| +| **"Context Driven Engineering"** | ⚠️ **Terminology error** | Perplexity search | ✅ Correct term: "Context Engineering" (Thoughtworks Tech Radar Vol 33, Nov 2025) | +| **"90%+ engineers at Alan"** | ✅ Yes | Emma Goldblum quote (article) | ✅ Verbatim exact | +| **"$20-200-1000/dev costs"** | ✅ Table present | Article | ⚠️ No methodology, 50x spread too large | +| **"Hugo 20+ years XP"** | ✅ Yes | Schema markup | ✅ Malt CTO 2012-2024, Writizzy founder 2025 | +| **"Published Feb 6, 2026"** | ✅ Yes | Metadata | ✅ Correct | +| **"Interviews 7 companies"** | ✅ List present | Article | ⚠️ No verbatims, no raw data | + +**Critical error detected**: Hugo miscites Thoughtworks framework as "Context Driven Engineering" when the actual term is "Context Engineering" (verified via Perplexity and Thoughtworks Technology Radar Vol 33). + +--- + +## Technical-Writer Challenge + +**Agent ID**: `ae2f481` (technical-writer subagent) + +**Challenge summary**: +- Initial score 4/5 **reduced to 2/5** after critical analysis +- Overestimated novelty — Spec/Plan/Act already in `spec-first.md`, QA/CI/CD in `production-safety.md` +- Underestimated marketing angle — no peer review, stats lack methodology, Writizzy link in footer +- Compared unfavorably to validated score-4 resources (Pat Cullen: 3 templates, Paddo: 10 actionable tips) + +**Legitimate points**: +- "Context Engineering" (Thoughtworks) is a real gap in the guide +- Corporate governance angle minimally covered +- Stats too vague for practical use ($20-1000 spread, no methodology) + +**Recommendation upheld**: Minimal integration (footnotes only), not full section. + +--- + +## Integration Decision + +**Action taken**: **Minimal integration** (2 footnotes) + +### 1. Context Engineering (Thoughtworks) — Priority HIGH + +**File**: `guide/methodologies.md` (after line 66, "Foundational Discipline" section) + +**Added**: +```markdown +> **Context Engineering**: Thoughtworks designates this broader approach "Context Engineering" +> in their Technology Radar (Nov 2025) — the systematic design of information provided to LLMs +> during inference. Three core techniques: context setup, context management for long-horizon +> tasks, and dynamic information retrieval. Related patterns in Claude Code: AGENTS.md, +> MCP Context7, Plan Mode. +``` + +**Rationale**: Legitimate framework gap, verified via Perplexity and Thoughtworks documentation. + +### 2. Corporate AI Marketplaces — Priority LOW + +**File**: `guide/adoption-approaches.md` (after line 277, "Larger Team" section) + +**Added**: +```markdown +> **Emerging approach**: Some organizations explore "corporate AI marketplaces" to pool AI +> skills, agents, and rules at the organizational level rather than individual teams +> (Hugo/Writizzy 2026). Few documented production implementations yet, but the concept +> addresses governance at scale. +``` + +**Rationale**: Interesting RH concept, minimal technical implementation details available. + +--- + +## Why NOT More Integration? + +### Rejected: Full "Team Governance" section + +**Reason**: Redundant with existing content: +- `guide/adoption-approaches.md` lines 236-278 already cover team coordination +- `guide/production-safety.md` covers hooks and permission rules +- `guide/security-hardening.md` covers team conventions + +### Rejected: Stats integration + +**Reason**: Unusable methodology: +- "$20-1000/dev" range is 50x spread +- No methodology documentation +- Our guide has more precise estimates (`ai-ecosystem.md`: $20-50 Claude Code typical) + +### Rejected: Citing "Context Driven Engineering" + +**Reason**: Term doesn't exist — Thoughtworks framework is "Context Engineering" + +--- + +## Comparison to Other Evaluations + +| Resource | Score | Templates/Code | Stats Quality | Integration | +|----------|-------|----------------|---------------|-------------| +| **Pat Cullen** (review-pr) | 4/5 | 3 templates | N/A | Full guide section | +| **Paddo Team Tips** | 4/5 | 0 (10 actionable tips) | N/A | Integrated throughout | +| **RTK** | 4/5 | 1 tool + examples | Measured 72.6% reduction | Full guide section | +| **Hugo AI Impact** | **2/5** | 0 | Vague ($20-1000 spread) | **2 footnotes only** | + +--- + +## Lessons Learned + +### Evaluation Process Improvements + +1. **Terminology verification**: Always cross-check framework names with authoritative sources (Perplexity, official docs) +2. **Gap analysis rigor**: Grep existing guide before claiming "missing content" +3. **Stats scrutiny**: Require methodology documentation, not just numbers +4. **Technical-writer challenge**: Proved valuable — caught overestimation of novelty + +### What Worked + +1. **Fact-check protocol**: Caught terminology error early +2. **Agent challenge**: technical-writer agent provided brutal but accurate reality check +3. **Minimal integration**: 10-minute footnotes vs 2-hour full section = better ROI + +--- + +## Sources + +- **Primary**: Hugo, ["AI's Impact on State of the Art in Software Engineering in 2026"](https://eventuallymaking.io/p/ai-s-impact-on-the-state-of-the-art-in-software-engineering-in-2026), Feb 6, 2026 +- **Verification**: Perplexity search for "Context Engineering Thoughtworks 2024 2025" +- **Authoritative**: [Thoughtworks Technology Radar Vol 33](https://www.thoughtworks.com/content/dam/thoughtworks/documents/radar/2025/11/tr_technology_radar_vol_33_en.pdf), Nov 2025 +- **Supporting**: [Thoughtworks macro trends blog](https://www.thoughtworks.com/insights/blog/technology-strategy/macro-trends-tech-industry-november-2025) + +--- + +## Metadata + +- **Evaluation date**: 2026-02-06 +- **Time spent**: ~45 minutes (research, fact-check, agent challenge, integration) +- **Agent tools used**: WebFetch, Grep, Read, Perplexity, Task (technical-writer) +- **Integration time**: 10 minutes (2 footnotes) +- **Files modified**: 2 (`guide/methodologies.md`, `guide/adoption-approaches.md`) diff --git a/guide/adoption-approaches.md b/guide/adoption-approaches.md index 54101c5..2297576 100644 --- a/guide/adoption-approaches.md +++ b/guide/adoption-approaches.md @@ -276,6 +276,10 @@ These are starting points, not rules. Team dynamics matter more than headcount. **Watch for**: Config drift. Without some coordination, setups diverge over time. Whether that matters depends on your team. +> **Emerging approach**: Some organizations explore "corporate AI marketplaces" to pool AI skills, agents, and rules at the organizational level rather than individual teams (Hugo/Writizzy 2026[^hugo2026]). Few documented production implementations yet, but the concept addresses governance at scale. + +[^hugo2026]: Hugo, ["AI's Impact on State of the Art in Software Engineering in 2026"](https://eventuallymaking.io/p/ai-s-impact-on-the-state-of-the-art-in-software-engineering-in-2026), Feb 6, 2026. Based on interviews with Doctolib, Malt, Alan, Google Cloud, Brevo, ManoMano, Ilek, Clever Cloud engineering teams. + --- ## Common Situations diff --git a/guide/methodologies.md b/guide/methodologies.md index 3c29a46..f27968b 100644 --- a/guide/methodologies.md +++ b/guide/methodologies.md @@ -65,6 +65,10 @@ Organized in a 6-tier pyramid from strategic orchestration down to optimization **Not just a feature (`/plan` command) — a systematic discipline.** +> **Context Engineering**: Thoughtworks designates this broader approach "Context Engineering" in their Technology Radar (Nov 2025)[^thoughtworks2025] — the systematic design of information provided to LLMs during inference. Three core techniques: context setup (minimal system prompts, few-shot examples), context management for long-horizon tasks (summarization, external memories, sub-agent architectures), and dynamic information retrieval (JIT context loading). Related patterns in Claude Code: AGENTS.md, MCP Context7, Plan Mode. + +[^thoughtworks2025]: Thoughtworks Technology Radar Vol 33, Nov 2025. [PDF](https://www.thoughtworks.com/content/dam/thoughtworks/documents/radar/2025/11/tr_technology_radar_vol_33_en.pdf). See also: [Macro trends blog post](https://www.thoughtworks.com/insights/blog/technology-strategy/macro-trends-tech-industry-november-2025). + **The Mental Model**: Planning isn't optional for complex tasks. It's the difference between: