The Arabic-first gateway to every leading AI model

FTBP AI Box puts a large catalogue of language, vision, image, speech and search models behind a single secure API, with smart routing, safety scanning and usage metering built in.

What is the AI Box?

A secure layer between your applications and the world's leading AI models. You integrate once. We handle access, safety, routing and billing.

Your applications Apps and websitesDeveloper tools and IDEsEnterprise systems 1 API FTBP AI Box Security and PII scanSmart routing + failoverMetering and budgetsArabic-first experience best fit Leading AI models
bolt

Integrate once

Teams stop wiring up a dozen vendors. One key, one invoice, one set of controls.

savings

Spend less, automatically

Easy requests go to fast, cheaper models. Hard ones go to frontier models.

verified_user

Safe by default

Sensitive data is caught before it leaves. Every request is metered and auditable.

Idea Build Check Deploy Operate FTBP AI Box: models, routing, safety and metering underneath every step
alt_route

factory

rule

deployed_code

account_tree

manage_search

chat

storefront

The value of the AI Box today

Three honest ways to measure what exists now. Reference figures use public list prices and are not revenue or contracted assets.

Catalogue breadth: models behind one API
Reference list-price value per month of a standard 2-billion-token workload across this catalogue
Estimated cost to rebuild the platform from scratch (gateway, routing, safety, metering, standards, operations)

Reference workload mix: 30% frontier, 50% mid-tier, 20% small models, at approximate public blended prices of $15, $3 and $0.40 per million tokens. The rebuild value is an engineering estimate. Both are planning references, not valuations.

Who it is for

Arabic-speaking and Gulf-region teams that want the best models without the integration, compliance and billing complexity.

code

Developers and AI startups

Fast access to every major model with one SDK and predictable cost.

storefront

SMEs and agencies

Add Arabic chat, content and automation to products without an AI team.

account_balance

Banks, telecom and enterprise

Governance, audit trails, spend control and vendor flexibility.

gavel

Government and semi-government

Regional hosting, policy controls and Arabic-first services.

school

Education and research

Shared, budgeted access for universities and learning platforms.

newspaper

Media and content

Arabic generation, translation, speech and moderation at scale.

Buyers we speak to

CTOs and heads of engineeringProduct and innovation leadsProcurement and compliance officersDigital transformation offices

Problems we solve

Arabic quality across modelsData protection and residencyMany vendors, many invoicesVendor lock-inRunaway AI spend

Production economics

A production-grade plan: highly available infrastructure, a real team and security built in. Planning estimates in US dollars, to be validated with vendor quotes.

Monthly infrastructure (production, HA)

Subscription plans (per month)

Core team (monthly)

Operations (monthly)

Year-one projection: monthly revenue vs cost

Customers ramp linearly over 12 months. Model-provider cost is assumed at 45% of revenue before any sponsor credits.

What we need from sponsors

Sponsorship turns a working platform into a contracted, supported, production service.

Cash sponsorship over 12 months (covers the base-case plan through year one)
Model credits per month (1.5x the reference workload, so growth never stalls)
SLACommitted rate limits, early access to new models, enterprise support channel
Co-brandJoint launch, case studies and regional events with the sponsor

Use of funds

The cash figure equals the base-case peak cumulative deficit plus a 25% buffer, and assumes no model credits. Credits would lower the cash needed. In the conservative scenario the deficit is higher, so we would slow hiring until revenue catches up.

1

2

3

4

5

6

7

8

help

help

help

help

What the box covers

Counts come straight from our live model registry, grouped by what each model can do.

Model catalogue

Every model available through the gateway, by family and capability.

Built for production teams

The platform around the models is what makes them usable at scale.

hub

One API

A single OpenAI-compatible endpoint. Switch models by changing a name, not your code.

alt_route

Smart routing

Requests are classified and sent to the right tier for quality, speed and cost, with automatic failover.

shield

Safety scanning

Prompts are screened for secrets and personal data before they leave the box.

visibility_off

Private by design

Clients see only our stable model names. Backend details and internal errors are never exposed.

query_stats

Metering and budgets

Per-client tokens, scopes, spend limits and a kill switch, with full usage reporting.

translate

Arabic first-class

Arabic-capable models and a bilingual product experience for the region.

Roadmap

Where we are and what sponsorship unlocks.

Today: working platform

Gateway, routing, scanning, metering and the catalogue above are built and running.

With sponsorship: contracted capacity

Move to provider agreements with guaranteed throughput, support and service levels.

Then: regional launch

Offer enterprise and developer plans with Arabic support and local data handling.

Partner with FTBP

We are looking for AI companies to sponsor model access and grow a regional AI gateway with us.

Get in touch