Spec Driven Development - The End Of Vibe Coding | enterJS 2026

Transcript

1. ENTERJS · 2026 Spec Driven Development The end of vibe coding. SPEAKER Daniel Sogl daniel.sogl@thinktecture.com Principal Consultant · Thinktecture AG
2. SPEAKER Daniel Sogl Principal Consultant at Thinktecture AG, focused on helping teams ship software faster and more reliably with AI. linktr.ee/daniel_sogl 01 · ROLE Principal Consultant Thinktecture AG — enterprise software across web, cloud & AI. 02 · FOCUS Developer Productivity & Gen AI Spec-driven workflows & agentic coding — so devs stop guessing. 03 · COMMUNITY MVP · Web Technologies Speaker, author & OSS contributor across the modern-web ecosystem. Spec Driven Development The End Of Vibe Coding 2 / 29
3. FROM THE FIELD Across 50+ client workshops in the last 4 years, requirements come up as the developers' bottleneck — almost every single time. Sources: Standish Group CHAOS Report · Engprax / Impact Engineering study (2024) · Wellingtone State of Project Management (2026) BEFORE WE BLAME THE AI Software rarely fails in the code. It fails in the requirements. 01 / PROBLEM #1 cause of failed & challenged projects — unclear or changing requirements 97% more likely to succeed with clear requirements up front 36% of organizations deliver their projects on time — the rest slip Spec Driven Development The End Of Vibe Coding 3 / 29
4. A human teammate pushes back on all three — they ask. Hold that thought. THE ANATOMY OF A BAD BRIEF Three ways requirements break. 02 / PROBLEM Incomplete Half the story. Edge cases, error paths, and the "what happens if…" nobody wrote down. // TODO: the rest Vague No number, no test — as many readings as there are readers. "make it fast" Contradictory Two truths, one codebase. The ticket says X, the stakeholder wants Y. spec ≠ ticket Spec Driven Development The End Of Vibe Coding 4 / 29
5. On real, unseen repos the top model drops from ~81% to ~46%. Measured: devs ran 19% slower with AI — while feeling 20% faster. 78% of teams saw more production incidents from AI code. Sources: Scale SEAL SWE-bench Pro Leaderboard (Jun 2026) · Stack Overflow Developer Survey 2025 · Veracode GenAI Code Security, Spring 2026 · METR developer- productivity study (2025) · New Relic State of AI Coding (2026) NOW HAND THAT TO A MACHINE AI agents fail more than you think. 03 / PROBLEM 59% best model on SWE-bench Pro (standardized scaffolding) 66% of devs' #1 AI frustration: code that's "almost right, but not quite" 45% of AI-generated code ships known vulnerabilities — flat since 2024 Spec Driven Development The End Of Vibe Coding 5 / 29
6. → → WHEN PROMPTS GO WRONG · LARBI ET AL. 2025 Mutate clean tasks into ambiguous, incomplete or contradictory ones → functional correctness drops sharply. Contradictory descriptions hurt most. THE CURE IS IN THE SAME DATA Let the agent ask and clarify, and accuracy climbs back to near full-spec levels. Structure beats guessing. Sources: Larbi et al., "When Prompts Go Wrong" (arXiv:2507.20439, 2025) · interactive-clarification studies on LLM code agents (2025–2026) THE AMPLIFIER AI doesn't fix vague requirements. It ships them — at machine speed. 04 / PROBLEM Ambiguous brief Incomplete · vague · contradictory Agent guesses No push-back. No clarifying question. Confidently wrong Compiles, looks done, solves the wrong problem. Spec Driven Development The End Of Vibe Coding 6 / 29
7. WHY AGENTS STUMBLE Five core limitations of AI coding agents. 05 / LIMITS 01 Short memory Forgets what you said two turns ago. 02 No trade-off sense Copies patterns without knowing why. 03 Ripple blindness Can't see the domino effect of a change. 04 Missing company rules Blind to your team's tribal knowledge. 05 Silent failures Fakes data, then reports "all tests green". Spec Driven Development The End Of Vibe Coding 7 / 29
8. WHAT IS A SPEC? A spec is to software what a blueprint is to a building. You wouldn't let the contractor decide the floor plan based on a hallway conversation. Yet that's exactly how most teams use AI coding agents today. WRITE THE BLUEPRINT. THEN HAND OVER THE TOOLS. Spec Driven Development The End Of Vibe Coding 8 / 29
9. DEFINITION What is Spec‑Driven Development? A development methodology where a detailed, living specification drives the entire process — especially the collaboration with AI coding agents. INTENT ARTIFACTS CODE 06 / DEFINITION 1 Spec first Write the specification before any code. The spec is the source of truth — not the diff. 2 AI-optimized Specs hand AI agents the full context upfront: goal, constraints, acceptance. 3 Iterative Specs evolve with the project. They are not throwaway requirements docs. 4 Consistency Uniform architecture across features — the constitution keeps agents in lane. Spec Driven Development The End Of Vibe Coding 9 / 29
10. Incomplete [NEEDS CLARIFICATION] The agent flags every gap instead of guessing. Vague Testable criteria GIVEN / WHEN / THEN + measurable FRs replace "make it fast". Contradictory Single source of truth Conflicts surface at the review gate — not in production. Same three failure modes — caught at the spec, before a single line of code exists. FROM PROBLEM TO CURE SDD heals the three wounds. 07 / THE FIX Spec Driven Development The End Of Vibe Coding 10 / 29
11. BACK TO THE FIVE LIMITS Complete context beats short memory · structured reasoning restores trade-off sense & ripple awareness · the constitution carries the company rules · validation gates end silent failures. HOW SPECS CHANGE THE GAME Four mechanisms — how specs improve AI. 08 / MECHANISMS MECHANISM 01 Complete context Goal, constraints & acceptance criteria — handed over up front, not pieced together each turn. MECHANISM 02 Structured reasoning Invariants, edge cases & error paths made explicit — less left to invent. MECHANISM 03 Validation checkpoints Gates between phases let you review intent, not just the diff. MECHANISM 04 Autonomous implementation Specs + tests become guardrails for hours of self-validating work. Spec Driven Development The End Of Vibe Coding 11 / 29
12. → → → WALKING THROUGH IT SDD in action — the four phases. 09 / PHASES 1 /specify Specify Write intent in plain prose. AI drafts; you approve. HUMAN + AI 2 /plan Plan AI derives architecture from spec and constraints. HUMAN + AI 3 /tasks Tasks Atomic units, ordered by dependency. AI 4 /implement Implement Agent executes, self- validates, checks off tasks. AI Spec Driven Development The End Of Vibe Coding 12 / 29
13. DETAILS SURFACE CLI · any editor OPINION Low — bring your agent SWEET SPOT New projects, new features DETAILS SURFACE CLI · any editor OPINION Medium — spec deltas track change SWEET SPOT Existing codebases (brownfield) DETAILS SURFACE Dedicated IDE OPINION High — full workflow SWEET SPOT Regulated · large teams PICK YOUR STACK SDD tooling landscape — three approaches, one goal. 10 / TOOLING 01 · CLI · OSS · 110K+★ GitHub Spec Kit Open-source CLI making specs executable artifacts. Agent- agnostic (Claude, Copilot, Cursor, Codex). Constitution → Specify → Plan → Tasks → Implement BEST FOR GREENFIELD 02 · FRAMEWORK · 55K+★ OpenSpec Universal planning layer for 25+ AI coding tools. Zero infrastructure — just Markdown in your repo. Propose → Apply → Archive BEST FOR BROWNFIELD 03 · IDE · AWS Kiro (AWS) Full IDE with SDD built into the editor. Turnkey workflow, enterprise guardrails. Requirements → Design → Tasks BEST FOR ENTERPRISE Spec Driven Development The End Of Vibe Coding 13 / 29
14. Everything lives in version control. No hidden state, no external SaaS — the repo is the system. SETUP Your repo, your rules — three files and you're running. 11 / SETUP 01 · PROJECT Constitution Project-level rules: tech stack, conventions, boundaries. Every agent reads it before touching code. ./CONSTITUTION.md 02 · PER-FEATURE Specs directory One Markdown file per feature. Plan, tasks, and generated artifacts live next to each spec. .specs/<feature>/spec.md 03 · AGENT Agent config Points the agent at the constitution and spec workflow. Pick the file your tool reads. CLAUDE.md · .cursorrules · AGENTS.md Spec Driven Development The End Of Vibe Coding 14 / 29
15. DESIGN & DEVELOPMENT Prove it works first — write acceptance tests before the feature. Speak the same language — business terms in code, specs, conversation. Contracts before UI — agree on the interface first, build the screen second. One thing, one responsibility — easy to change, hard to break. ARCHITECTURE & QUALITY One team owns one feature — end-to-end, no hand-offs. Security is a default — not a feature request, baked in from day one. Explicit "done" criteria — each phase has a gate, no shortcuts. Track your building blocks — every dependency pinned and auditable. MANDATORY WORKFLOW Specify → Contract → Plan → Tasks → Red → Green → Refactor → Review CONSTITUTION Boundaries before code — the repo constitution. 12 / CONSTITUTION Spec Driven Development The End Of Vibe Coding 15 / 29
16. A spec is a structured Markdown document that captures the three pillars every AI agent needs to succeed — stripping these down causes silent drift. ANATOMY What makes a good specification? 13 / ANATOMY PILLAR 01 Intent = "What are we building and why?" The goal, the user story, and the motivation behind the feature. # Feature · User story # Motivation PILLAR 02 Constraints = "What rules must we follow?" Tech choices, patterns, and boundaries that prevent the agent from drifting off course. ## Tech stack ## Must / Must-not PILLAR 03 Acceptance criteria = "How do we know it's done?" Concrete, testable conditions — the agent checks its own work against these. GIVEN / WHEN / THEN FR-001 · FR-002 · ... Spec Driven Development The End Of Vibe Coding 16 / 29
17. DO Be specific about behavior and edge cases. Define acceptance criteria as testable scenarios. List constraints — especially what not to do. Reference existing patterns in the project. Include error handling expectations. DON'T Describe implementation — let the plan phase derive it. Write vague requirements like "make it fast". Assume the agent knows your codebase or history. Mix multiple features in one spec. Skip non-functional requirements (limits, audit). CRAFT Writing good specs — do's & don'ts. 14 / CRAFT Spec Driven Development The End Of Vibe Coding 17 / 29
18. The Spec You write intent — AI drafts the spec, flags ambiguity FR-004: Evaluation latency MUST NOT exceed [NEEDS CLARIFICATION: target p99 latency?] 3 15 / SPEC [NEEDS CLARIFICATION] AI flags what it cannot assume — no silent guessing 1 Input 2 Draft Flag Spec Driven Development The End Of Vibe Coding 18 / 29
19. The Plan You steer tech direction — AI derives the architecture research.md — tech choices, trade-offs data-model.md — entities & relations contracts/ — OpenAPI spec quickstart.md — dev setup REVIEW GATE You approve plan + artifacts before tasks are generated. 3 16 / PLAN SUPPORTING ARTIFACTS + REVIEW GATE GENERATED ALONGSIDE PLAN.MD You are the gatekeeper at every phase 1 Input 2 Plan Gate Spec Driven Development The End Of Vibe Coding 19 / 29
20. The Tasks AI breaks the plan into atomic, verifiable units T01 Database Schema & Migrations FILE migrations/001_flags.sql DEPENDS — ACCEPTANCE Schema matches data-model.md, audit table append-only VALIDATE npm run migrate && npm run test:schema T02 [P] FlagRepository FILE src/repositories/flag.repository.ts DEPENDS T01 ACCEPTANCE Contract tests pass, archival works, no hard delete VALIDATE npm run test:repo T03 [P] FlagService — Rollout Eval FILE src/services/flag.service.ts DEPENDS — ACCEPTANCE Hash distribution within ±2% of target over 10k samples VALIDATE npm run test:service T04 REST Controller (OpenAPI-bound) FILE src/controllers/flag.controller.ts DEPENDS T02, T03 ACCEPTANCE All endpoints match contracts/openapi.yaml VALIDATE npm run test:contract && npm run test:e2e 2 17 / TASKS SAMPLE TASKS 1 Run Tasks [P] = parallelizable. Agent runs independent tasks concurrently. Spec Driven Development The End Of Vibe Coding 20 / 29
21. Implementation /implement — AI executes, validates, retries 12 / 12 Agent respects dependencies, parallelizes independent tasks, and checks off tasks.md as it progresses. 47 / 47 Self-validation after each task. On failure the agent revisits the task — it does not skip ahead. ~35 min From approved spec to a running service with audit log, cache, and rollout logic. Human review time not included. 0 lines of code hand-written. 100% reviewable trail: spec → plan → tasks → commits. 18 / IMPLEMENT ⌗ TASKS COMPLETED ✓ TESTS GREEN ◷ END-TO-END Spec Driven Development The End Of Vibe Coding 21 / 29
22. Recap — what just happened Spec → Plan → Tasks → Code YOU + AI Spec spec.md · scenarios + FRs Your intent, AI-drafted, you reviewed → YOU + AI Plan plan.md + 4 artifacts Your tech direction, AI-derived → AI Tasks 12 atomic units AI-generated from plan → AI Code ~1,800 LOC + tests AI-implemented, self- validated Your job shifts from writing code to shaping intent and reviewing artifacts at every gate. 19 / RECAP Spec Driven Development The End Of Vibe Coding 22 / 29
23. Specify → Clarify → Plan → Tasks → Analyze → Implement Both are review gates, not code — they catch ambiguity and drift while fixes are still cheap. HARDEN THE LOOP Two more gates — /clarify & /analyze. 20 / REFINEMENT AFTER /specify · BEFORE /plan /clarify Structured, coverage-based questions surface under-specified areas. Answers are captured in a Clarifications section — the agent stops guessing before the plan exists. CUTS DOWNSTREAM REWORK AFTER /tasks · BEFORE /implement /analyze Cross-artifact consistency & coverage check across spec, plan & tasks. Flags gaps, contradictions & constitution violations — read-only, never edits. LAST GATE BEFORE CODE Spec Driven Development The End Of Vibe Coding 23 / 29
24. Spec Kit, OpenSpec & Kiro all ship the spec as files in the repo. Perfect for one dev on one repo — fragile the moment real teams and org structure show up. SDD is the right direction — but the execution isn't production-ready yet. That gap is the opportunity. MY HONEST TAKE Great approach — the tooling isn't ready yet. 21 / REALITY Developer-locked Specs live in Git. PMs, QA & domain experts can't co-author or review where they actually work. No spec sync Nothing reconciles specs across features, teams, or services — they silently drift apart. Conflicting specs When two specs contradict, who wins? No arbiter — just a merge conflict. Teams & poly-repo Two devs on one spec? One spec across many repos? Largely unsolved today. Spec Driven Development The End Of Vibe Coding 24 / 29
25. USE SDD WHEN SKIP SDD WHEN RULE OF THUMB Two or more prompts to explain? Write a spec. One sentence? Skip it. Thoughtworks Technology Radar: Spec-driven development placed in "Assess" — many SDD frameworks suit greenfield, OpenSpec targets brownfield. BOUNDARIES The right tool for the right job. 22 / USE VS. SKIP Greenfield projects Spec becomes single source of truth from day one. Complex features Contracts across services, schemas, and domains need structure. Multi-agent coordination Shared specs are the coordination primitive. Compliance & audit Spec + review + commit = audit trail for free. Small bug fixes Spec overhead exceeds the change. Rapid prototyping Specs need clarity — exploration doesn't. UI-heavy / creative Pixel choices live in Figma, not Markdown. One-off scripts Throwaway code — just prompt it. Spec Driven Development The End Of Vibe Coding 25 / 29
26. TAKE HOME Five takeaways. 23 / TAKEAWAYS 01 Projects fail in the requirements, not the code. Incomplete, vague & contradictory briefs — and AI ships those flaws at machine speed. 02 Intent as truth — not code as truth. Specs drive everything downstream: plan, tasks, code, tests, review. 03 Specifications become executable — through AI. From passive docs to active source of working software. 04 Choose your tool — Spec Kit · OpenSpec · Kiro. Greenfield · Brownfield · Enterprise — pick once, keep the workflow consistent. 05 Industry convergence on structured intent. SDD is becoming standard practice — GitHub, AWS, and the open-source community are all heading here. Spec Driven Development The End Of Vibe Coding 26 / 29
27. Exceptional at pattern completion, not at mind reading. DEN DELIMARSKY GITHUB PRINCIPAL PM " Spec Driven Development The End Of Vibe Coding 27 / 29
28. PROJECT & REQUIREMENTS FAILURE Standish Group — CHAOS Report standishgroup.com Engprax — Impact Engineering: requirements & success (2024) engprax.com Wellingtone — State of Project Management (2026) wellingtone.co.uk REQUIREMENTS MEET AI Larbi et al. — "When Prompts Go Wrong" (2025) arxiv.org/abs/2507.20439 Survey — Code Generation with LLM-based Agents (2025) arxiv.org/abs/2508.00083 SDD PRACTICE & TOOLING Thoughtworks — Radar: Spec-driven development thoughtworks.com/radar AI CODING RELIABILITY Scale SEAL — SWE-bench Pro Leaderboard scale.com/leaderboard Stack Overflow — 2025 Developer Survey (AI) survey.stackoverflow.co/2025 Veracode — GenAI Code Security (Spring 2026) veracode.com/blog METR — AI & Experienced Dev Productivity (2025) metr.org New Relic — State of AI Coding (2026) newrelic.com TOOLS GitHub Spec Kit OpenSpec Kiro CHECK IT YOURSELF Sources & further reading. 24 / SOURCES Spec Driven Development The End Of Vibe Coding 28 / 29
29. Q & A · THANK YOU Thanks for your attention. Let's build software that actually does what it says. Daniel Sogl daniel.sogl@thinktecture.com Principal Consultant · Thinktecture AG linktr.ee/daniel_sogl