Pay for what your visitors use

The pricing logic: pay as you go

Traditional assistant tools mostly sell fixed, bloated monthly packages; you pay the same whether you use them or not. RAQS flips that. The core charge is usage-based: the more a visitor speaks in a voice session (the voice minutes processed), the more it costs. A quiet day means a low usage bill.

The reason is where cost actually arises. Each voice session runs realtime audio transport (LiveKit), speech-to-text and text-to-speech (Azure Speech), and a language model (Azure OpenAI gpt-4o-mini). These operations consume real resources per minute. Pay-as-you-go reflects that real cost transparently; it never penalizes you for capacity you didn't use.

In practice this means a small blog and a busy e-commerce store aren't forced into the same mold. As your traffic grows, your payment grows proportionally; when a season ends and visitors drop, the usage charge falls on its own. Voice assistant pricing follows the rhythm of your campaigns and seasons.

What the base plan includes

The base plan is the part that keeps the service continuously available regardless of usage. Your site connection, your knowledge base and your dashboard stay running even when no one speaks. The usage charge is only added on top of real voice conversations.

The plan covers going live with one line of code, managing the knowledge base (crawl, file upload, paste text/FAQ), the allowed-domains allowlist, mascot/persona selection, and conversation logs with analytics. In the multi-tenant architecture, each customer's data is isolated with row-level security (RLS); this isolation and security layer is standard on every plan.

On the security side, an SSRF-guarded crawler, an origin allowlist and Cloudflare Turnstile anti-abuse protection are included as well. The infrastructure is hosted in Azure's EU region and is designed with KVKK and GDPR in mind. These are not add-on modules — they are part of the core service.

How quotas and limits work per plan

Each plan carries two kinds of limits: usage quotas and capacity limits. On the usage side, a plan may include an amount of voice minutes; once you exceed that included amount, pay-as-you-go billing kicks in. So your plan's usage is consumed first, and anything beyond is usage-based.

On the capacity side, knowledge base size scales with the plan. Site crawling works with a per-plan page quota: larger plans crawl more pages and build a deeper knowledge base. The crawl respects robots.txt; JS-rendered SPA pages are processed with a headless browser and clean text is extracted with trafilatura. A bigger site requires a higher page quota.

These two axes are independent: a content-heavy but rarely-spoken-to site may see a high page quota but low voice-minute usage; conversely, a thin but heavily visited campaign page may produce low page quota with high minute usage. When choosing a plan, you should weigh both axes.

An illustrative monthly picture (qualitative)

To give a concrete feel, consider two typical cases — not numbers, but the logic. First case: a small service site. You pay the base plan fee; visitors ask only a few hundred short voice questions a month. Here the usage charge stays small; most of the bill is the predictable base plan. The cost is predictable and low.

Second case: a busy e-commerce store. You pay the same base plan, but visitors talk much more, asking 'is this in stock, how much, show me.' Because voice-minute usage rises, the usage charge is added noticeably on top of the base plan. But since those conversations serve sales directly, the fee scales with the value it brings.

What matters is the pattern: the base plan stays fixed and predictable; the part on top grows only with real conversation volume. Because you can track usage (voice minutes, session counts, traffic) from the dashboard, there are no surprises and you can adjust the plan accordingly.

Start with the free trial

Before committing to anything, you start with a free trial. You connect your site, crawl the knowledge base, pick a mascot and persona, and try the assistant by speaking to it like a real visitor. The goal is to see — with your own eyes, before paying — how accurately RAQS answers from your content and how it changes the feel of your site.

RAQS runs as a managed cloud service: you don't set up servers, host models, or wrestle with scaling. You only add one line of code and use the dashboard. At the end of the trial you move to the base plan + pay-as-you-go model; because you've already seen what usage profile your real traffic produces, you choose the right plan deliberately.

<script async src="https://raqs.ai/v1/raqs.js"
  data-raqs="YOUR_SITE_KEY"></script>