AI product studio

AI products, built end to end.

Blox Labs designs, builds and ships AI products — from the model layer to the launch site. I run two of my own: Playcut and Outfeed AI.

Founded
2025
Studio
Toronto
Team
One person
Live products
02
  • Google Veo
  • Google Imagen
  • Google Gemini
  • xAI Grok
  • fal.ai
  • Anthropic Claude
  • Model Context Protocol (MCP)
  • React
  • Vite
  • Astro
  • Tailwind CSS
  • shadcn/ui
  • TypeScript
  • NestJS
  • Node.js
  • Hatchet
  • Clerk
  • Stripe
  • MongoDB
  • PostgreSQL
  • Google Cloud Storage
  • Cloudflare
  • Docker
  • pnpm
  • MDX / Astro Content Collections
  • Google Search Console API

01 Selected work

Products I build, and run.

Two live AI products I designed, built and still operate — plus work built for clients.

Own product · Live

01 / 02

Playcut

AI video & image studioSince 2025

A multi-model AI studio for cinematic video and images.

  • 01 Routes every prompt to the best model for the shot
  • 02 AI actors that keep the same face from shot to shot
  • 03 Every model on one credit balance
Woman walking through a neon-lit city street at nightNeon night
Editorial portrait of a woman in a wood-panelled libraryEditorial light
Sunglasses under a pink and blue gradient lightGradient glow
Woman walking through a field of tall grass at golden hourGolden hour
Sample outputs from Playcut

Own product · Live

02 / 02

Outfeed AI

AI social media managerSince 2026

A chat-first social media manager for nine platforms.

  • 01 Say what to post — no dashboard to learn
  • 02 A preview for every platform before it ships
  • 03 Publishes to nine networks from one conversation
Illustration · The Outfeed AI chat flow

02 Capabilities

What I build for you.

Five kinds of work — every one already shipped in a product I run.

  1. 01 AI product, built end to end

    An idea to a live AI product — model layer, backend, app, billing, launch site.

    You have a product in mind that needs AI at its core. I build the whole thing: the model orchestration, the API, the app, the job pipeline that makes it reliable, the billing, and the marketing site it launches on. One person, one stack, no handoffs between agencies.

    What you get

    • Working product in your repo, not a prototype
    • Model layer that routes each task to the right provider
    • Backend API and background job pipeline
    • Web app frontend
    • Auth, billing and usage metering
    • Deployment and infrastructure setup

    Shipped in Playcut ↗ Outfeed AI ↗

  2. 02 Multi-model AI orchestration

    Route each task to the model that's actually best at it — not whichever one you wired up first.

    Most AI features wrap a single model and inherit all of its weaknesses. I build the routing layer instead: per-task model selection, one model picker instead of a tab per tool, one credit balance across providers, and a catalog you can extend when a better model ships.

    What you get

    • Per-task model routing with a documented catalog
    • Provider abstraction so a model swap isn't a rewrite
    • One credit balance across every provider

    Shipped in Playcut ↗

  3. 03 Agent tooling & MCP servers

    Make your product something an AI agent can actually drive.

    An MCP server turns your app into a set of tools Claude and other agents can call directly. I build the server, design the tool surface so an agent can succeed without hand-holding, and handle auth and rate limiting.

    What you get

    • MCP server exposing your product's real capabilities
    • Tool surface designed for agent success, not API completeness
    • Auth, scoping and rate limiting
    • Docs an agent can read

    Shipped in Playcut ↗

  4. 04 SaaS platform engineering

    The unglamorous layer that decides whether a product survives real users.

    Multi-tenant auth, workspaces and sharing rules, background jobs that retry properly, media pipelines, subscription billing and usage limits. This is the work that turns a demo into something you can charge for.

    What you get

    • Multi-tenant auth and organization model
    • Workspace, folder and sharing permissions
    • Durable background job pipeline with retries and timeouts
    • Subscription billing, plan limits and usage metering

    Shipped in Outfeed AI ↗ Playcut ↗

  5. 05 Launch site & content infrastructure

    A fast static marketing site with an SEO system behind it, not just pages.

    A static Astro site that scores well and loads instantly, plus the infrastructure underneath: schema markup generated from validated data, a content pipeline, an anti-cannibalization keyword index, and verified index-status tracking against the Search Console API.

    What you get

    • Static Astro site — fast, accessible, zero unnecessary JS
    • JSON-LD generated from data, so markup can't drift from the page
    • MDX content pipeline with validated frontmatter
    • Keyword index and campaign planning system
    • Search Console integration for real index-status tracking

    Shipped in Playcut ↗ Outfeed AI ↗

03 The studio

One person. No handoffs.

Blox Labs is one person. I design the product, write the backend, build the frontend, run the infrastructure, write the copy, and answer the support email. That is unusual to say out loud, and it is the point: there is no account manager between you and the person who wrote the code.

Mehran G., founder of Blox Labs Inc.
Mehran G. Founder & CEO · Blox Labs in 𝕏
  • 01

    You talk to the builder

    Every message, call and commit comes from the person writing the code. There is no account manager in between.

  • 02

    Proof, not promises

    Every capability on this site points at a product that is live today — you can go and use it.

  • 03

    One stack, end to end

    Model layer, backend, app, billing and the launch site, built by one person — so nothing gets lost between vendors.

Start a project

Have an AI product to build?

Tell me what you're making. You'll hear back from the person who would build it.

Start a project