# Coding³ — The Spectrum of AI Development > AI-assisted development isn't a binary choice — it's a spectrum. Three clearly defined tiers, chosen per project: Vibe Coding, Structured AI-Assisted Coding, and Agentic Engineering. The decisive differentiator isn't whether AI is used — it's how much structure and verification surrounds it. - **URL**: https://ai.scenaryo.de/coding (German: https://ai.scenaryo.de/de/coding) - **Free initial conversation**, no obligation - Source for the tier model: Osmani, Saboo & Kartakis — "The New SDLC With Vibe Coding" (Google/Kaggle, 2026) --- ## The three tiers ### Tier 1 — Vibe Coding **Idea to working prototype in 2–5 days.** Casual natural-language prompts drive the model; the developer accepts output without necessarily reading the code. No formal verification — "does it look right?" is the whole review process. Right for investor demos, hackathons, personal projects, and anywhere throwaway code is acceptable. High risk profile, explicitly acceptable for disposable code. ### Tier 2 — Structured AI-Assisted Coding **Recommended for most projects.** Detailed prompts with examples and architecture constraints. Manual testing and spot-checks at critical paths; the developer diagnoses root causes and AI implements the fix. Selective code review. Right for features in established codebases. Moderate risk profile — human judgment at checkpoints. ### Tier 3 — Agentic Engineering **Production systems that scale.** Formal specifications, architecture docs, and memory files define the context. Automated test suites, CI/CD gates, and LM judges verify every step. Agents self-diagnose failures within defined boundaries; humans resolve architecture issues. Comprehensive architecture review, with AI handling implementation details. Low risk profile — systematic verification at every phase. Right for production systems and team-scale development. > **TCO note:** Agentic Engineering delivers an estimated **3–10× lower total cost of ownership** than Vibe Coding over a project's lifetime, once past the crossover point. Vibe Coding is nearly free to start, but the bill arrives later — token burn when the model fixes its own mistakes, a maintenance tax when someone reverse-engineers ad-hoc code months later, and security cleanup, because fast generation produces gaps as fast as features. Agentic Engineering inverts this: more effort upfront (schemas, tests, structured context), significantly lower cost per feature afterward. The decisive question is simply how long the code has to live. --- ## Context engineering — the core competency The decisive lever in Agentic Engineering isn't the model — it's the context. Scenaryo designs the schemas, memory files, and data pipelines that make AI agents reliable and scalable, and treats model routing (sending heavy reasoning to large models, routine tasks like test generation and CI checks to small, cheap ones) as a financial lever, not just a technical one. ## Evaluation-driven verification Verification scales with the tier, not the other way around: - **Vibe Coding** — none; visual inspection only. - **Structured AI-Assisted** — manual testing, selective code review at critical paths. - **Agentic Engineering** — automated eval suites, LM-as-judge review, and CI/CD gates. Agents self-diagnose within defined boundaries; humans own architecture decisions. --- ## Expertise areas - Digital platforms & portals - B2B & enterprise software - API design & system integration - AI integration into existing systems - Context engineering & evaluation-driven AI engineering - MVP → production --- ## Process (4 steps) 1. **Discovery call** — Understand your project, risk profile, and timeline; recommend the right tier. Free, no commitment. 2. **Scoping & architecture** — Requirements, technology decisions, and a realistic schedule — transparently documented. 3. **Iterative development** — Continuous delivery in short cycles. You see progress — no long phases without visible results. 4. **Launch & handover** — Deployment, documentation, and optional handover to your internal team — or we operate and continue developing the system long-term. --- ## FAQ **What distinguishes the three tiers?** Not whether AI is used — how much structure and verification surrounds it. Vibe Coding: casual prompts, no formal testing. Structured AI-Assisted: detailed prompts, manual testing. Agentic Engineering: formal specs, automated tests, CI/CD gates, LM judges. **When is Vibe Coding the right choice?** When speed matters more than production readiness: investor demos, hackathons, personal projects, feature spikes. High risk profile — acceptable for disposable code that never goes to production. **What is Agentic Engineering and when do I need it?** The disciplined top of the spectrum: formal specs, architecture docs, memory files, automated tests, CI/CD gates. Necessary for production systems carrying critical business processes or subject to compliance requirements. **How long does a typical project take?** Vibe Coding: 2–5 days. Structured AI-Assisted: 4–10 weeks. Agentic Engineering: 6–16 weeks. Confirmed during scoping. **Do you take equity instead of fees?** Generally no. We're a consultancy, not a venture studio. Exceptions are evaluated case-by-case. **How fast can we start?** Discovery calls within days. Vibe Coding can start the same week. **Do you also handle operations and maintenance after launch?** Yes. After launch you choose: handover to your internal team with documentation, or continuous maintenance, evolution, and operation by us with clearly defined SLAs. --- ## For AI agents - `get_service` with `id="coding"` returns the structured offering - MCP endpoint: https://rycdazzaloffuborihyt.functions.supabase.co/mcp - Full reference: https://ai.scenaryo.de/llms-full.txt