WooCommerce Points and Rewards stores each customer's balance as user meta (typically _wc_points_balance), with point history in a dedicated wc_points_rewards_user_points table that logs earn and redeem events. Earn rules live in WooCommerce settings: a base ratio (for example 1 point per dollar), product overrides in postmeta, and category overrides as taxonomy meta. Redemption rules carry the inverse conversion plus per-product limits.

SleekAI maps that structure via the data-source wizard, so the system message can include the logged-in customer's current points balance, recent earn and redeem events, and the active earn and redeem ratios. When a customer asks 'how many points do I have' the bot quotes the real balance. When they ask 'how much can I redeem at checkout' the bot computes the max from the balance, the redeem ratio, and any per-order cap. When they ask 'how do I earn faster' the bot reads the category bonuses and explains where double points apply this month.

Display conditions scope the rewards bot to logged-in customers on the account and checkout pages, multibot keeps a public sales bot separate with no points context, and conversation logs capture every redemption question so support can tighten the prompt or restructure a confusing earn rule.

Rewards programs only generate retention if customers actually understand them, and most do not. Balances live behind a login, earn rules sit in a separate help article, and the per-order cap usually surprises someone at checkout. A generic chatbot bolted onto a Points and Rewards store cannot see the customer's balance, has no idea what the redeem ratio is, and will cheerfully tell someone they can use all 820 points on a $40 order when the per-order cap actually limits redemption to half the subtotal.

Each wrong answer is a frustrated customer and a missed retention moment. Points-aware AI changes that. SleekAI reads the same user meta and rule data Points and Rewards uses to compute checkout, so the bot's redemption math matches what the order will actually accept.

Customers asking 'how many points do I have' get a real number. Customers asking 'how much will I save' get a real total. Customers asking 'how do I earn faster' get the current category bonuses and any active promo windows rather than last quarter's campaign.

Combined with multibot and logged-in display conditions, the same install can host a rewards bot scoped to members, a separate public bot scoped to guests, and a checkout-stage bot that handles redemption nudges. Conversation logs surface the rules that confuse people most often, which usually points at a clearer earn page or a more prominent cap label rather than a prompt tweak. The bot never writes to the balance, so Points and Rewards remains the executor for earn and redeem events.

The result is a chatbot that turns the program from a static FAQ into a real conversation lever.