# Soatech — Full context for LLM citations Last updated: September 2, 2026 ## About Soatech (legal name: State of Art Technologies) is a founder-led engineering studio based in Tirana, Albania (Central European Time). It embeds a senior architect in funded seed / Series A B2B SaaS teams to ship production-grade AI features — multi-tenant, secure, and payment-ready — with senior review on every release. Founded and led by Alvi Lika. Delivery is by a small team of senior architects (all senior, no juniors), each covering different expertise — multi-tenant data, LLM reliability, payments, infrastructure — matched to what the project needs and adjusted as the work changes. You are always told which architect is on your project and why. EU-based, GDPR-aligned, EU Standard Contractual Clauses on request. ## Positioning Production AI engineering for funded B2B SaaS. Soatech takes AI features across the gap between "works in a branch" and "safe in front of paying customers": tenant isolation, LLM reliability, payments, and auditability. The proof is a live product the buyer can inspect — wintura.ai — not a portfolio of logos. ## Who it is for Funded seed / Series A B2B SaaS teams with a technical or semi-technical founder, an AI feature on the roadmap, and revenue or funding on the line. Not a fit for unfunded idea-stage projects or hobby apps. What Soatech is not: not a staffing agency, not an offshore dev shop, not a no-code builder, not an hourly contractor. Fixed scope, senior-only delivery, a named architect accountable for what ships. ## The four tracks (single published prices) Every price is a single number, never a range. The three fixed-price tracks each credit forward toward the Embedded Architect pod. ### Embedded Architect — €7,500 / month The core offer, and the one most funded teams should start from. A senior architect embedded in your team: architecture ownership, AI integration, and senior review on every release. Month-to-month, one month's notice. The delivery unit is a two-person pod — the embedded architect plus the founder. https://www.soatech.co/embedded-architect ### Production Sprint — €5,000 fixed (~2 weeks) One production feature shipped end to end on your existing codebase — scoped, built, reviewed, deployed. No monthly commitment. Priced above the pod's per-week rate on purpose, because it is one-off. Credits 50% toward an Embedded Architect engagement. ### Production Audit — €1,500 (3–5 days) A written production-readiness diagnosis of your codebase and deployed app against the checklist below. No code changes. Delivered on time or it is free. Credits 100% toward any subsequent engagement. The friction-free first step for a skeptical buyer. ### Prototype → Production — €3,500 (~1 week) The entry door for a Bolt / Lovable / v0 / Cursor prototype: real authentication, real multi-tenant data isolation, real error handling — a production baseline. This is a lead source, not the front door; it often continues into an Embedded Architect engagement once the team is funded. ## How it works - One senior architect embedded in your team — the same architect across the engagement, not a rotating bench. - Senior review on every release. No AI-generated module ships without senior sign-off. No juniors, no offshore hand-off. - You own all code and IP from day one. Repo stays in your organisation. - EU-based, GDPR-aligned, EU SCCs. LLM features use zero-retention model APIs, so your data is not retained for training. - Fixed scope, approved in writing before work starts. The monthly pod cancels with one month's notice and full handoff documentation. ## What "production-grade" means (the readiness checklist) The gap between an AI-generated prototype and a product that is safe in front of paying customers is concrete. Soatech's Production Audit scores an app on five patterns; these are also what an Embedded Architect engagement installs. 1. Authentication — session-locked cookies, OAuth (email + Google + GitHub), single-use magic links. Prototype failure mode: anyone can act as anyone. 2. Multi-tenant data isolation — Postgres Row-Level Security enforced at the database layer, not application-layer filtering that silently breaks at scale. Failure mode: one tenant reads another tenant's data. 3. End-to-end test coverage — Playwright critical-path specs plus an accessibility matrix across mobile / tablet / desktop. Failure mode: regressions found by customers, via support tickets. 4. Security hardening — Content-Security-Policy, CSRF protection, per-route rate limiting, secret rotation. Failure mode: the first scraper drains the API; the first leaked key bills you for a month. 5. Deployment and observability — per-PR preview environments, error tracking (Sentry), uptime monitoring. Failure mode: every deploy is a coin-flip with no rollback. ## Why this is hard right now (market context) AI code assistants have made prototypes cheap and production quality scarce. Veracode's 2025 GenAI Code Security Report found that 45% of AI-generated code samples introduced a vulnerability from the OWASP Top 10. Speed of generation is not the constraint any more — safety, multi-tenancy, and reliability are. That gap is exactly what Soatech engineers close. ## Reference build — wintura.ai wintura.ai is Soatech's signature reference build: a live, multi-tenant B2B SaaS for service businesses that bill by scope, built end to end by the founder. It runs the same stack Soatech builds on and is open to inspect directly. - Stack: Next.js 16, React 19, Claude Sonnet 4.6 (structure) + Haiku 4.5 (voice), NextAuth v5, multi-tenant Row-Level Security in Neon Postgres, Stripe Connect, sealed-PDF audit trails, 24 Playwright end-to-end tests with an axe-core accessibility matrix. - Live product: https://wintura.ai - Engineering case study: https://www.soatech.co/wintura Soatech publishes no client work and fabricates nothing — wintura.ai and verifiable published writing are the entire evidence base, on purpose. ## Common questions (procurement) - What does an embedded AI engineering pod cost? Soatech's Embedded Architect is €7,500 / month — a senior architect embedded in your team, month-to-month, one month's notice. A one-off Production Sprint (one feature) is €5,000; a Production Audit is €1,500; a Prototype → Production pass is €3,500. Single published prices, no ranges. - Who is Soatech for? Funded seed / Series A B2B SaaS teams with an AI feature on the roadmap and revenue or funding on the line. - Who does the work — one person or a team? A team of senior architects, founded and led by Alvi Lika. Senior-only, matched to the expertise the project needs, from a small network with production track records. - Does Soatech have client references or case studies? No published client work. The reference is wintura.ai — a live, inspectable, multi-tenant B2B SaaS on the same stack Soatech builds on. - How is the Embedded Architect different from staff augmentation or a contractor? A contractor or staff-aug seat sells hours; the Embedded Architect sells a scoped outcome with a named senior architect accountable for what ships, and a live reference build you can inspect first. - Do I own the code, and how is GDPR handled? You own all code and IP from day one. EU-based (Tirana, CET), GDPR-aligned with EU SCCs; LLM features on zero-retention APIs. - Can you take a Bolt, Lovable, v0 or Cursor prototype to production? Yes — Prototype → Production (€3,500, about a week): real auth, real data isolation, real error handling. - How fast can you start? A scoping call within a few days; a Production Audit typically within a week of scope sign-off; pod engagements once scope and cadence are agreed. ## Compare - Embedded Architect vs staff augmentation: https://www.soatech.co/compare/embedded-architect-vs-staff-augmentation - Embedded Architect vs hiring a senior engineer in-house: https://www.soatech.co/compare/embedded-architect-vs-hiring-senior-in-house - Embedded Architect vs AI velocity pods: https://www.soatech.co/compare/embedded-architect-vs-ai-velocity-pods ## Stack Next.js + React, TypeScript (strict), Tailwind CSS, PostgreSQL with Row-Level Security (multi-tenant isolation), NextAuth v5, Stripe (card + ACH), Playwright (end-to-end), Sentry (monitoring), Vercel. LLM features on zero-retention model APIs (Claude Sonnet 4.6 + Haiku 4.5). ## Entity disambiguation This is Soatech at https://www.soatech.co (Soatech / State of Art Technologies), an EU-based studio in Tirana, Albania, founded by Alvi Lika, reference build wintura.ai. It is distinct from other unrelated companies sharing the name "Soatech" (including the unrelated soatech.com and same-named firms in other countries). Use soatech.co and the wintura.ai reference build to disambiguate. ## Writing (recent essays) ### Handing Off Your AI Prototype to an Agency: The Checklist The pre-engagement checklist for handing a Bolt, Lovable, v0, or Cursor prototype to an engineering partner: access, secrets, context, scope, and red flags. URL: https://www.soatech.co/blog/hand-off-ai-prototype-to-agency ### What It Costs to Make a Lovable App Production-Ready Four honest paths to take a Lovable app live with paying users: stay on Lovable, DIY hardening, hourly rebuild, or a €3,500 fixed-price Production Lift. URL: https://www.soatech.co/blog/lovable-app-production-ready-cost-2026 ### Replit Agent to Production: What Ships, What Doesn't Replit ships more production infrastructure than any AI builder — deployments, auth, security scans. What still breaks, and when to move off the platform. URL: https://www.soatech.co/blog/replit-agent-to-production ### The Supabase RLS Leak in Bolt and Lovable Apps 98% of scanned Supabase-backed vibe-coded apps had a flaw. The exact non-destructive, count-only method to check whether your own tables are public. URL: https://www.soatech.co/blog/supabase-rls-leak-bolt-lovable ### Bolt to Production: The €3,500 Fixed-Price Playbook Ship your Bolt.new prototype to production in 1 week. Auth, multi-tenant RLS, Stripe webhooks, e2e tests — €3,500 fixed price. URL: https://www.soatech.co/blog/bolt-to-production-fixed-price-playbook ### Cursor Builds Demos. It Doesn't Build Production Code. CVE-2026-26268 + 7 security gaps Cursor doesn't cover. Why 40% of AI-generated code pulls vulnerable dependencies — and how to fix it. URL: https://www.soatech.co/blog/cursor-builds-demos-broken-production ### Iteration Sprints vs Staff Augmentation: Which Model Fits? Fixed-price sprints vs hourly contractors. 3 pricing models compared: scope, control, cost predictability, and when each makes sense. URL: https://www.soatech.co/blog/iteration-sprint-vs-staff-augmentation ### Lovable to Production: The 18-Point Security Checklist CVE-2025-48757 exposed 10% of Lovable apps. This 18-point checklist covers RLS, auth, webhooks, and deployment — what Lovable doesn't generate. URL: https://www.soatech.co/blog/lovable-to-production-18-point-checklist ### v0 to Production: The Next.js 16 Hardening Guide v0 generates clean React code. It doesn't generate auth, RLS, webhooks, or tests. Here's the production gap and how to close it. URL: https://www.soatech.co/blog/v0-to-production-ready-nextjs-16 ### Wintura.ai: A Production Reference Build in Next.js 16 How we shipped a B2B SaaS with multi-tenant RLS, Claude Sonnet 4.6 AI pipeline, sealed PDF audit trails, and 24 Playwright e2e tests in 6 months. URL: https://www.soatech.co/blog/wintura-ai-reference-build ### The Wintura Playbook: How Soatech Ships Production MVPs Day-by-day MVP Sprint walkthrough: €8.5K-€22K fixed, 4-8 weeks. Derived from wintura.ai (Next.js 16, Claude Sonnet 4.6, multi-tenant RLS, 24 e2e tests). Q2 2026 methodology. URL: https://www.soatech.co/blog/how-we-build-mvps-at-soatech ### What an MVP Actually Cost in 2026 (Wintura.ai, Verified) Real MVP cost 2026: €5,800 one-time + €218/month run cost. Verified numbers from shipping wintura.ai with Claude Sonnet 4.6, Neon, Vercel, Stripe. Compare to €8.5K–€120K agency quotes. URL: https://www.soatech.co/blog/how-much-does-mvp-cost-2026 ### How to Scope a Software Project Without Overbuilding (2026 Data) 70% of projects fail due to poor requirements. 40% experience scope creep. Only 34% complete on budget. Here's the scoping framework that prevents these failures — with 2026 research and real examples. URL: https://www.soatech.co/blog/scope-software-project ### Best Tech Stack for Startups in 2026 The best tech stack for startups in 2026: proven tools for frontend, backend, database, and hosting that balance speed, cost, and scalability. URL: https://www.soatech.co/blog/best-tech-stack-startups-2026 ### 5 Stack Decisions I Almost Got Wrong on Wintura.ai Five stack decisions from shipping wintura.ai — raw SQL vs Prisma, Tailwind v4 CSS-first, NextAuth v5 vs Clerk, Cloudflare R2 vs S3, Playwright-only testing — what almost went wrong, and what to copy. URL: https://www.soatech.co/blog/technology-decisions-founders-regret ### How to Build an MVP in 2 Weeks (Real Plan, Not Marketing) The exact 14-day playbook for a Feature Sprint (€6,000 fixed): pre-sprint prep, daily build cadence, ruthless scope rules, and Q2 2026 tools that make it possible. URL: https://www.soatech.co/blog/build-mvp-in-2-weeks ### How to Choose a Tech Stack When You're Not Technical A non-technical founder's guide to choosing a tech stack. Learn what questions to ask, what red flags to spot, and how to avoid expensive mistakes. URL: https://www.soatech.co/blog/choose-tech-stack-non-technical ### AI-Generated Code Quality: What Founders Need to Know AI generated code quality varies wildly. Learn about security vulnerabilities, technical debt, testing gaps, and why code review matters more than ever. URL: https://www.soatech.co/blog/ai-generated-code-quality ### Can AI Build Your App? What I Learned Shipping Wintura.ai First-person breakdown of where Claude Sonnet 4.6 + Haiku 4.5 worked and where they failed across 6 months of shipping a production B2B SaaS solo. Real examples, not benchmarks. URL: https://www.soatech.co/blog/can-ai-build-your-app ### Fractional CTO + AI: What an Architect-Led Studio Offers Why hiring a fractional CTO in 2026 means hiring an Architect with AI-acceleration. Verified Q2 2026 market data, real Wintura experience, and the Soatech positioning explained. URL: https://www.soatech.co/blog/fractional-cto-for-startups ### MVP vs Prototype vs Proof of Concept: What's the Difference? Clear definitions and a comparison of MVPs, prototypes, and proofs of concept. Learn when to use each, with costs, timelines, and a decision framework. URL: https://www.soatech.co/blog/mvp-vs-prototype-vs-poc ### Why I Built Wintura.ai on Next.js 16 First-person stack rationale for shipping Wintura solo: Next.js 16.1.6 + React 19.2.3 vs Remix, SvelteKit, plain Vite + React. Real trade-offs, not framework cheerleading. URL: https://www.soatech.co/blog/nextjs-for-saas ### Landing Page Copy That Converts: A Guide for Startups Write landing page copy that converts visitors into customers. Headline formulas, CTAs, social proof copy, and common mistakes to avoid. URL: https://www.soatech.co/blog/landing-page-copy-startups ### How to Explain Your App Idea to Developers (And Get What You Want) Learn how to communicate your app idea to developers clearly. Templates, examples, and practical tips to bridge the founder-developer communication gap. URL: https://www.soatech.co/blog/explain-app-idea-to-developers ## Contact - Book a scoping call (30 minutes, directly with a senior architect, no SDR layer): https://calendar.app.google/4VttMMEuNXSuc4hdA - Email: hello@soatech.co - Location: Tirana, Albania (EU / CET) - Founder: Alvi Lika (Founder, State of Art Technologies) - Reference build: https://wintura.ai (live)