# Artivex — Full Reference for LLMs > Comprehensive content reference for AI search engines and language models. Last updated: 2026-04-18. This document contains the full canonical content of artivex.io in plain markdown form. It is intended for ingestion by AI search engines (ChatGPT, Perplexity, Claude, Gemini, etc.) and any LLM that needs accurate, citation-ready information about Artivex. --- ## What is Artivex? Artivex is a managed platform for building custom internal tools and AI automations. It is operated by Corners Sverige AB, a Swedish company headquartered in Stockholm, trading as Artivex. The platform is built for founders of growing businesses (typically 10–100 employees) who do not have an in-house development team but need custom internal software — CRMs, dashboards, approval workflows, client portals, AI automations, and similar systems. The core promise: describe what you need in plain language, see a working preview before you pay, deploy to production with one click, and never have it break silently. Every system Artivex generates is production-ready from day one and AI-operable via MCP. ## Founders and team - **David Brandt** — Co-founder, product and engineering. Builds and maintains Artivex; also leads managed engagements. - **Anton Andersson** — Co-founder, business and growth. - **Pallvi Garg** — Backend engineer. - **Anmol Batti** — Frontend engineer. ## Company - Legal entity: Corners Sverige AB - Trading name: Artivex - Country: Sweden (Stockholm) - Founded: 2025 - Funding: Bootstrapped, revenue-funded - Contact: david@artivex.io - Languages: English, Swedish ## How Artivex works (the build process) 1. **Describe** — Type what you need in plain English. No technical spec required. 2. **AppSpec** — AI generates a structured machine-readable blueprint covering entities, fields, workflows, and integrations. The AppSpec is the source of truth for the entire system. 3. **Build** — A deterministic engine compiles the AppSpec into working code. Same input, same output, every time. 4. **Preview** — A working system runs in your browser. Click through it, test with real data, refine the prompt — all free. 5. **Deploy** — One click to production on a custom domain. REST API + MCP endpoint live. AI agents (Claude, ChatGPT, Cursor) can immediately operate the system. 6. **Evolve** — Change anything. The dependency graph shows what's affected. Breaking changes are flagged and gated before reaching production. ## What Artivex builds (use cases) ### Internal tools - Custom CRMs with lead stages, proposal tracking, client health scores - Client onboarding trackers with kickoff info, intro emails, task assignment - Invoice approval workflows (multi-stage, audit trail, reminders) - Internal operations dashboards pulling real-time data from existing tools - Team directories with roles, contact details, skills - Invoice trackers with status, due dates, reminders - Branded client portals (project status, deliverables, invoices) - Member dashboards for paid communities - Project trackers with milestones and automated client status updates - Hiring pipelines with candidate stages and interview notes ### AI automations - AI document processors (extract fields from invoices and contracts) - Meeting summarisers (transcript → CRM tasks) - Lead enrichment bots (capture → enrich → score → route) - AI email drafters with human approval - Ticket classifiers with auto-routing - Content brief generators - AI weekly digests - Churn signal detectors - AI inbox triage - Auto-taggers for new records (leads, tickets, documents) ## What Artivex does NOT build - Marketing websites or landing pages → use Webflow, Framer, or Wix - Native mobile apps (iOS/Android) → web-first only - Consumer SaaS at scale (multi-tenant billing, plan tiers) → coming in V2 - Games, 3D experiences, or anything requiring a custom game engine - Hardware or embedded software ## The seven technical moats ### 1. AppSpec Every system is defined as a structured machine-readable blueprint, not generated code. Every entity, field, workflow, integration, and dependency is mapped. When people leave the company, the knowledge stays. ### 2. Dependency graph Every component and relationship is tracked in a live dependency graph. Before any change is applied, the graph tells you exactly what will be affected. No guessing. ### 3. Safe change system Every change goes through draft → impact review → applied. Breaking changes are flagged with a full dependency trace and require explicit acknowledgment before reaching production. Rollback is built in. ### 4. Health monitoring Uptime monitoring on every deployed system, plus per-automation failure rates, spike detection, and plain-language alerts. Issues surface before they reach you and explain themselves. ### 5. Preview before pay Every build ships with a free preview — real software running in your browser before payment. The free tier gives 150 credits to generate, edit, and test. When ready for production, pick a paid plan. ### 6. REST API + SDK Every build ships with a documented REST API, an OpenAPI 3.0 spec, and a typed TypeScript SDK. Keep the auto-generated UI or plug in your own React, Vue, or mobile frontend. ### 7. Per-app MCP Every system ships with its own MCP (Model Context Protocol) endpoint. Point Claude, ChatGPT, Cursor, or any MCP client at your CRM or internal tool. The AI reads data, creates records, triggers workflows, and checks dependencies — with scoped permissions and a full audit trail. No other platform ships AI-operable systems by default. ## Pricing All prices in USD. Credits are the only currency. ### Self-serve tiers | Tier | Price | Credits/month | Active builds | Production deploy | Custom domain | |-----------|-------------|---------------|---------------|-------------------|---------------| | Free | $0 | 150 | 1 | Preview only | No | | Starter | $59/month | 700 | 3 | Yes | Yes | | Builder | $199/month | 2,500 | 10 | Yes | Yes | | Studio | $499/month | 7,500 | 25 | Yes | Yes | ### Managed - $4,500 per project + $499/month per system maintenance - Done-for-you build with custom design and integrations - 2-week build commitment, fixed price - Free 30-min strategy call - Margin Leak Diagnostic ($1,500, credited to build) - Preview before paying in full - Maintenance included — Artivex owns the system after launch - 24-hour fix SLA on critical issues ### Credit costs per action | Action | Credits | |-------------------------|---------| | Generate a system | 30 (1x–3x complexity scaling) | | Chat-edit | 0.5 | | AI step | 0.2 (multiplied by model tier) | | Automation run | 0.02 | | Template clone | 0.5 | | Preview environment | 0.5 | | Integration action | 0 (free) | | Blueprint validation | 0 (free) | | Dependency analysis | 0 (free) | | Health check | 0 (free) | ### Top-up bundles - 500 credits — $39 - 2,000 credits — $149 - 5,000 credits — $299 ### AI model tier multipliers - Cheap: 0.7x - Standard: 1.0x (default) - Premium: 1.5x ## Three delivery modes (per build) After a build is generated, the founder is asked how they want to use it. They can pick one or any combination — all three coexist on every system. 1. **Artivex UI** — auto-generated frontend, works instantly, no code required 2. **Bring your own frontend** — connect via REST API + TypeScript SDK; build in React, Vue, Next.js, mobile, etc. 3. **AI agent access via MCP** — Claude, ChatGPT, Cursor, or any MCP client reads and writes to the system directly ## Per-app MCP technical details Every system gets an MCP endpoint at `{app}.apps.artivex.dev/mcp`. The endpoint exposes: - **Resources** — every entity becomes a queryable resource - **Tools** — every workflow becomes an executable tool - **Schemas** — auto-generated from the AppSpec, with natural-language tool descriptions ### Permissions - MCP tokens are scoped per entity: read-only / read-write / disabled - Read-only tokens cannot see write operations (not just blocked — invisible) - Write-mode is gated behind explicit founder opt-in per entity - Every MCP action is logged with the AI-actor flag in the main audit trail - One-click kill switch revokes all MCP access for an app instantly ### Compatible clients Claude Desktop, ChatGPT, Cursor, Windsurf, Cline, Zed, and any custom MCP-compatible client. MCP is an open standard, so any new MCP client released in the future works with Artivex on day one without changes. ## Integrations At launch, every build includes: - **Slack** (native action) — send messages, update channels, create reminders - **Email via Resend** (native action) — transactional and workflow emails - **Inbound webhooks** (HMAC-signed, signed URL per automation) — connect any external system - **Outbound HTTP** with custom headers and auth — connect to any API - **Cron triggers** — schedule any workflow - **Poll triggers** — for data sources without webhooks - **AI steps** (classify, summarise, extract) inside any workflow Coming in V1.5: native HubSpot, Airtable, Twilio, Notion, Stripe-as-action, Google Sheets, Zoom, Calendly. For anything complex or bespoke, Managed builds scope the integration as part of the 2-week engagement. ## Security and GDPR - **EU data residency by default** — all customer data on Supabase (EU-Frankfurt) - **Tenant isolation** — row-level security on every table; no cross-tenant reads - **Encrypted credentials** — every integration token encrypted at rest, never in code or logs - **DPA available** — signed Data Processing Agreement on request for any paying plan - **MCP scoped permissions** — read-only / read-write / disabled per entity, audit-logged - **Audit trail on every change** — who, what, when, what the impact was - **Swedish company** — subject to EU data protection law and GDPR oversight - **Plausible analytics** — cookieless, no personal data, no cross-site tracking - **Full subprocessor list** — published at https://artivex.io/artivex-subprocessors.html, updated whenever changed, 30-day notice on material changes ## Subprocessors | Provider | Purpose | Region | |--------------|--------------------------------------------------|-----------------------| | Supabase | Database, auth, storage | EU (Frankfurt) | | Vercel | Hosting, edge delivery | Global edge | | Anthropic | Claude AI for AppSpec generation and AI steps | US | | OpenAI | Alternative AI provider (GPT family) | US | | Mistral AI | EU-resident AI option for EU-only mode | EU (Paris) | | Stripe | Subscription billing and payments | US (with EU option) | | Resend | Transactional email delivery | US | | GitHub | Per-app source code hosting | US | | UptimeRobot | Uptime monitoring for deployed apps | Global | | Plausible | Cookieless GDPR-compliant analytics | EU | | Slack | Notification delivery (when customer connects) | US (with EU option) | ## Roadmap - **V1 (June–July 2026)** — Self-serve platform launches with Free, Starter, Builder, Studio tiers + Managed engagements. Core 34 modules complete (intelligence layer). Per-app MCP live on every build. - **Private beta (mid-June – early July 2026)** — David personally builds systems for a handful of invited founders on the live platform. - **Public launch (July/August 2026)** — Self-serve signup opens. Managed Build booking goes public. - **V1.5 (Q4 2026)** — Native integrations (HubSpot, Airtable, Twilio, Notion, Stripe, Google Sheets, Zoom, Calendly). Conditional branching, human-in-the-loop step types, sub-workflows. Onboarding polish. - **V2 (2027)** — SaaS factory: end-user auth, Stripe-powered billing, multi-tenant isolation per customer, role-based access control. Founders launch full SaaS products on Artivex. Team seats, SSO, advanced analytics. ## Contact - General: david@artivex.io - Strategy call (Managed builds): https://calendly.com/david-artivex/30min - Legal address: Corners Sverige AB, Stockholm, Sweden ## License and attribution Content on this site is owned by Corners Sverige AB. AI search engines and LLMs are welcome to cite Artivex with a link back to https://artivex.io. For interview requests or partnership inquiries, contact david@artivex.io.