Setup & tiers

The AI Assistant is a subscription add-on. Whether it is switched on, which AI provider powers it, and how many tokens you get are all controlled centrally — you don't manage API keys inside the CRM. This page explains what those settings mean and how usage is measured.

Enabling the assistant

AI is turned on as part of your Xelynx subscription and managed from portal.xelynx.com. When it is enabled, the AI Assistant appears in the sidebar and is ready to use. If it is not part of your plan, requests return a clear message asking you to enable it from the client portal. The provider, key and token allowance sync into the CRM automatically, so a change made in the portal takes effect within a couple of minutes — there is nothing to configure on the CRM side.

Provider and model

The assistant can run on one of two providers — Anthropic (the default) or DeepSeek — chosen for your account. To keep costs down, the CRM automatically picks the right model for each request: lightweight questions such as simple lookups, counts and status checks use a fast, economical model, while write actions, chart requests and deeper analysis (forecasts, comparisons, full reports) use a more capable one. This routing is automatic and needs no input from you.

Token allowance and usage

Usage is metered in tokens — the unit of text the AI reads and writes. Your subscription includes a monthly token limit, shown as a meter in the assistant header (for example 2.4M / 10M tokens). The bar turns orange as you approach the limit. Every request counts toward the total, including the data the AI reads to answer you, so broad analytical questions use more than a quick count.

Good to know

The token limit is shared across your whole team and all chat sessions, not per user. Usage is tracked against your account centrally, so the figure in the CRM matches what the portal reports. If you regularly hit the ceiling, increase your token tier from the client portal rather than changing anything in the CRM.