Verto AI
PROJECT
OVERVIEW
Verto AI is a full-stack creative workspace built on Next.js 16 that combines three distinct subsystems: an 8-agent LangGraph pipeline for presentation generation, an Inngest-powered mobile design generator, and a hosted MCP server that exposes presentation workflows to external AI clients. The architecture deliberately separates these subsystems while sharing a common AI runtime router and Prisma data layer.
Verto AI is an interconnected platform using LangGraph, Inngest, and MCP to programmatically generate layout-aware presentations — serving both human users through a visual editor and AI agents through a protocol-standard MCP endpoint.
“Layout-first generation guarantees agents are structurally aware of target formats before writing a single word — content fits the visual container, not the reverse.”
SYSTEM
ARCHITECTURE
Engineered for production durability, type safety, and scalable domain isolation. Every module operates with strict boundaries and predictable failure handling.
MODULAR DOMAIN ISOLATION — BUILT FOR ZERO-DOWNTIME DEPLOYMENTS, TYPE-SAFE CONTRACTS, AND FAIL-SAFE EXCEPTION BOUNDARIES.
PROJECT
IMPACT
Production performance measured under live traffic load, latency stress profiles, and automated system profiling.
ALL OUTCOMES ARE EMPIRICALLY VERIFIED IN PRODUCTION ENVIRONMENTS — MEASURED UNDER HIGH CONCURRENCY, PEAK LATENCY PROFILES, AND AUTOMATED SYSTEM HEALTH AUDITS.
MCP Tools
LangGraph Agents
Architecture Decision Records
AI Providers (BYOK)
KEY
CAPABILITIES
Core capabilities engineered for high-scale reliability, intuitive operator workflows, and real-time production execution.
PRODUCTION-TESTED SYSTEM FEATURES — ARCHITECTED FOR LOW-LATENCY INTERACTION, DETERMINISTIC STATE HANDLING, AND COMPREHENSIVE TELEMETRY.
8-Agent LangGraph State Machine
A sequential pipeline of 8 specialized agents: projectInitializer creates the DB record, outlineGenerator produces slide outlines, layoutSelector assigns layout types per slide, contentWriter generates structure-aware copy, imageQueryGenerator creates search terms, imageFetcher loops until all slides have images, jsonCompiler assembles the final Slide[] JSON, and databasePersister saves everything. Recursion limit: 150.
Layout-First Content Generation (ADR-005)
Layout selection happens as step 3, before content writing at step 4. The content writer receives the chosen layout type and generates copy that fits the target visual structure. This prevents the common AI problem where generated text overflows or underflows slide containers.
Hosted MCP Server (11 Tools, 4 Resources)
Exposes presentation lifecycle operations to external AI clients over Streamable HTTP (protocol version 2025-03-26). Supports both remote Bearer-token auth for hosted use and local stdio transport for development. Session-based — clients must initialize before sending tool calls.
BYOK Multi-Provider Runtime
Users store API keys for Google, OpenAI, and Groq with validation metadata. Every generation request passes through a business-rule router: free tier → hosted default, BYOK eligible → resolve preferred provider and model, fallback → safe default. The same resolver is used by web generation, streamable API, MCP tools, and mobile design jobs.
Recursive Slide Editor (ADR-006)
Slides are stored as recursive JSON content trees. The same tree structure renders identically across the editor, preview, public share view, and PDF export. Zustand manages the editor state with theme switching, content editing, and the recursive rendering engine.
DEVELOPMENT
PROCESS
A rigorous, phased engineering lifecycle designed to transform architectural requirements into scalable, production-ready systems.
DISCIPLINED SYSTEM METHODOLOGY — ARCHITECTED FOR END-TO-END OBSERVABILITY, DETERMINISTIC REVIEWS, AND ZERO-REGRESSION RELEASE CYCLES.
TECHNOLOGY
STACK
An engineered architectural map outlining foundational nodes, runtime environments, and type-safe deployment frameworks.
DETERMINISTIC ARCHITECTURE STACK — PROFILED FOR PRODUCTION LATENCY, STRICT TYPE SAFETY, AND HIGH-CONCURRENCY SCALABILITY.