LEGAL

Fair use policy.

Version 1.1-draft · 25 October 2026

This policy forms part of the Client Services Agreement. It explains how usage is measured against a package’s monthly allowances and what the services may not be used for.

1. How usage is measured

• Chatbot conversation: one visitor session on the client’s chatbot. A conversation ends after 30 minutes without a message (configurable per client between 5 and 240 minutes); a later message starts a new conversation. A conversation is counted once, when the assistant first replies. Conversations that fail because of a system error are not counted.
• Workflow execution: one successful run of an automation template for one event (for example a form submission, a pipeline stage change or a support request), or one scheduled run that produced work (for example the weekly report). Each event is identified by an event id: a re-delivered or retried event with the same id is counted once, and failed runs are not counted.
• AI voice minute: the length of each AI-handled call from answer to hang-up, rounded up to the next whole minute per call, from the voice provider’s records.
• SMS message: each message sent or received through the client’s Kevara-managed number, counted once regardless of length. Outbound messages that fail or are undelivered are not counted.
• CRM integration: each external system connected to the client’s automations (for example a CRM, booking tool or spreadsheet).
• AI model usage is tracked (as estimated tokens) to monitor costs; it is not billed separately.
• Usage is counted per billing cycle starting on the subscription renewal date, from provider records and Kevara’s logs, which are durable and deduplicated.

2. Allowances, notifications and limits

• Kevara notifies the client at 80% (warning), 100% (allowance used up) and 150% (escalation) of each allowance.
• Overage charges apply only after the client has authorised overage billing at rates disclosed in advance; the default maximum is $50 per billing cycle.
• At 150% of an allowance, or when the authorised overage cap is reached, further chargeable usage is suspended unless the client authorises a higher limit. For chatbots this means new conversations receive a short message with the business’s contact details; for workflows, new events are refused until the limit is raised or the cycle renews.
• Services a package does not include (for example voice or SMS on Starter) are not available until the package is upgraded.

3. Technical limits

• Chat: up to 20 messages per minute per visitor connection and 600 per hour per client website; up to 30 exchanges per conversation; 1,000 characters per message.
• Client events API: up to 120 events per minute per client.
• Knowledge base: up to 400 sections per client.
• Limits protect all clients and may be adjusted; Kevara gives at least 30 days’ notice of any material change to this policy or its limits, except where an immediate change is needed to stop abuse or protect the service.

4. Not permitted

• Unsolicited bulk messages (spam), messages or calls without the consent the law requires, or ignoring opt-outs.
• Robocalls or AI calls to people who have not consented, or calls without the required disclosures.
• Unlawful, deceptive, discriminatory, harassing or infringing content, or impersonating others.
• Processing health information covered by HIPAA, payment card numbers, government identification numbers or other sensitive data through the chatbot or workflows, unless agreed in writing.
• Using the services to give professional legal, medical or financial advice.
• Attempting to bypass usage limits, access another client’s data, overload the services, or reverse-engineer them.

5. Enforcement

Kevara may limit or suspend the affected feature, after notice where practical, to stop a breach of this policy, protect people or other clients, or comply with law, as described in the Client Services Agreement. Questions: t.thomas@kevarallc.com.