Free to start. Grows with your audience.

Yes. The Starter plan is free with no credit card required. You can integrate the SDK, go live with your assistant, and start receiving revenue events at no cost. The Starter plan covers up to 100,000 conversation turns per month, which is sufficient for most early-stage publishers.

If your volume exceeds the Starter limit, you will be notified with at least 72 hours before we pause matching on your account. Upgrading to Studio resumes matching immediately.

Revenue share is applied to the gross CPM earned by your placements. On the Starter plan, you keep 70% of the gross CPM. On Studio, you keep 75%. Platform terms are negotiated individually and may exceed 75% at sufficient volume.

Example: if your assistant generates 1,000 placement impressions in a month at a $3.45 average CPM, the gross revenue is $3.45. Your Starter share is $2.41 (70%). Your Studio share would be $2.59 (75%). Velocity's take is the remainder.

A conversation turn is one complete exchange in an AI assistant conversation: one user message plus the assistant's reply. Each turn is a potential candidate for contextual matching, though Velocity only serves one placement per full conversation session regardless of turn count.

The turn count is measured at the point of intent evaluation, not at the point of ad delivery. A session with 12 turns counts as 12 turns toward your monthly limit, even if only one placement is served.

Starter publishers receive monthly payouts on a net-30 schedule. This means earnings from January are paid by the end of February, allowing time for advertiser billing cycles to close. The minimum payout threshold is $50. If your balance is below $50 at the end of the month, it rolls forward.

Studio publishers receive bi-weekly payouts. Platform publishers can negotiate weekly payout schedules. All payments are made via ACH bank transfer (US publishers) or international wire (international publishers). Paypal and checks are not supported.

Yes. You have full control over category exclusions. From the publisher dashboard you can permanently exclude any topic category (finance, health, political content, adult content, competitor categories, etc.). Exclusions take effect within 5 minutes and apply across all placements served to your assistant.

Studio and Platform publishers can also set keyword-level exclusions for more granular control. Exclusions reduce eligible fill rate, so we recommend reviewing the revenue impact estimate shown in the dashboard before enabling multiple exclusions simultaneously.

The Velocity SDK is framework-agnostic. It exposes a single async function that accepts the conversation thread as a string array and returns a placement object (or null if no match). You integrate the placement rendering into your existing response rendering layer, which works regardless of whether you use LangChain, Vercel AI SDK, OpenAI streaming, Anthropic streaming, or a custom implementation.

Server-side and client-side integrations are both supported. The SDK has Node.js, Python, and browser ESM builds. Streaming-compatible with SSE and WebSocket response patterns.