Welcome to ToRun

ToRun is an AI-native platform that brings the world's best language, image, audio, video and reasoning models together behind one capability-first interface. Whether you are a founder shipping a product, a creator monetizing your audience, or a business automating workflows, ToRun gives you the same orchestration layer the largest AI teams use - without the lock-in, without manual provider management, and without paying retail prices on every call.

Our Mission

We believe AI should be a craft, not a dependency.

Today, most AI products force you to pick a single vendor, accept their pricing, hand over your data, and pray they do not deprecate the model you built around. ToRun rejects that posture. We are building a platform where you own your keys, own your data, own your workflows, and pay only for what the router cannot already cover with your existing subscriptions.

Our mission is to make capability-first AI orchestration the default for builders, creators and businesses worldwide - with verifiable model routing, real-time billing accuracy, and a marketplace economy that rewards the people who actually build the workflows that work.

The Problem We Solve

Every team building with AI today runs into the same five walls:

1. Vendor lock-in. You picked OpenAI, then Anthropic released a better Claude. You picked Anthropic, then Google released Gemini 2 Pro for half the price. Switching costs real engineering time and breaks production.
2. Hidden pricing. Provider invoices are opaque. You see one line item per month. There is no way to attribute cost to a specific feature, a specific user, or a specific call. Refunds and audits are nightmares.
3. No portability. Your conversation history, your memory, your tool definitions - they all live in someone else's database. If you leave, you start from zero.
4. Capability mismatch. Modes that need vision use models without it. Modes that need tools use models that strip tool-calling. The provider's 'best model' is rarely the right one for your specific call.
5. No monetization layer. If you build a great workflow, you can sell it on a download site as a ZIP file or build a whole SaaS around it. There is no middle ground.

ToRun solves all five.

What We Build

Four product pillars, one orchestration core.

1. Chat - Multi-Provider Conversation

Mode-first conversation across 16 product surfaces (chat, code, canvas, research, council, deep-thought, voice, image-edit, video-gen, translator, and more) and 17 capabilities (vision, tool-calling, structured-output, embeddings, TTS, STT, realtime audio, and more). Bring your own key (BYOK) for any provider, or let ToRun route to the cheapest model that satisfies your capability requirements. Every call is billed atomically with a full pricing snapshot - no surprises at month-end.

2. Workflow + Marketplace - Sell Your Automations

A custom thin-DAG engine that turns natural-language prompts into shippable automations. Build once, publish to the ToRun Marketplace, and earn up to 75% revenue share when customers subscribe. Creators retain ownership; ToRun handles billing, distribution, sub-processor compliance, and customer support escalation.

3. Persona / Social / Creator - The Civitai x TikTok x Patreon Hybrid

Build an AI persona with a public profile, post AI-generated content, attract subscribers, accept tips, sign brand deals, and stream live. Personas have their own monetization stack: subscriptions, one-off tips, brand-deal disclosures, live-stream tipping, and engagement-based payouts. Phase 1 gives you a clean storefront; Phase 2 opens the keyword stock-exchange so your persona can compete for premium ad real-estate.

4. bouncyad.com Ad Exchange - Consumer-Side

ToRun is the consumer side of a Vickrey second-price keyword auction. Surfaces in the platform participate transparently - every impression carries a verifiable provenance trail, every click is logged with the model that selected it, and every conversion goes back to the right advertiser without any third-party tracking pixel. Coming H2 2026.

How ToRun Works

A request enters ToRun and runs through this pipeline:

1. Intent classification - a self-hosted ONNX model identifies what you are trying to do, with zero external cost.
2. Capability resolution - your mode declares the capability set required (for example 'chat + vision + tool-calling'). The router filters AiModel rows by those capabilities.
3. Routing decision - the router ranks surviving candidates by AiModelPricing (the canonical price source - never quoted off AiModel directly), respecting your BYOK preferences, plan tier, and budget rules.
4. Execution - the chosen provider executes the call. We capture token counts, output, latency, and any tool-call results.
5. Billing - a BillingRecord is written atomically with the full pricing snapshot (PricedAtRate, FundSource, ByokTierApplied, CostBreakdownJson). Every AI call lands exactly one billing row, regardless of success, partial completion, or failure.
6. Outbox dispatch - domain events go through OutboxMessage in the same DB transaction as the state change, then async to a Kafka-compatible event bus for downstream consumers. Real-time summaries are fed by event-driven consumers, never end-of-day batches.

This architecture means: cost is always attributable, history is always portable, routing is always verifiable, and a single broken provider never takes down the platform.

Our Values

Capability-first. We do not pick favourite vendors. We pick the right tool for the call, every time. The router is the product.

Real-time everything. No batched billing, no end-of-day summaries, no 'we will have your numbers tomorrow.' Every counter you see is current to the last few seconds.

Verifiable trust. Audit chains are hash-linked. Billing records are immutable. Pricing snapshots survive provider price changes. If you have a dispute, you have the receipts.

Provider-agnostic by design. We support OpenAI, Anthropic, Google, xAI, Mistral, DeepSeek, Together, RunPod, self-hosted models, and the next dozen providers that have not launched yet. Switching takes minutes, not months.

Open economics. Marketplace creators see exactly what we take and exactly what they earn. Advertisers see exactly what the surface owner gets. No black boxes between the call and the revenue split.

Our Team

ToRun is shipped by a small, dense team focused on capability-first architecture, real-time billing accuracy, and verifiable model routing. We are remote-first with primary operations in Istanbul, Turkey, and engineers across Europe and Asia.

Founders have shipped distributed systems at scale, run production AI infrastructure for Fortune 500 customers, and bootstrapped two creator-economy products before this one. We are heads-down on Phase 1 launch and intentionally not raising capital until our ARR clears the next milestone.

Trust + Security

• GDPR / KVKK / CCPA compliant from day one. Data Subject Requests have a 30-day SLA, all handled through an in-app workflow with full audit trail.
• SOC 2 preparation in progress; certification target Q3 2026.
• CSAM scanning via PhotoDNA runs at every tier - no bypass, ever.
• Penetration testing by an independent firm every six months.
• At-rest encryption for BYOK secrets, billing data, and audit logs via cloud key-management service.
• In-transit encryption for every API request via TLS 1.3.
• Hash-chained audit logs for billing, moderation, and admin actions - tamper detection runs nightly with operator alerts.

Open Standards, No Walled Gardens

ToRun does not lock you to a payment processor, a model provider, a CDN, or a storage vendor. Paddle is our merchant of record today, but the billing pipeline is provider-agnostic. Bunny.net runs storage and CDN, but the abstraction layer can swap. Every long-term decision is made with portability in mind - if a future tenant wants to export every row of their workspace, our DataSubjectRequest workflow gives them a portable ZIP.

Get Started

Open a free account at torun.ai and explore the platform. Free tier includes generous limits on chat, image, and translation; paid plans unlock higher capabilities, marketplace access, and persona monetization.

Have questions? See our pricing page, browse the persona directory, or reach our team via contact.

Welcome to ToRun.