Rules tell the Claimlane AI Agent how your organization handles different customer requests. They define principles, constraints, and typical actions without forcing every case through a fixed script.
Before you start
Decide which ticket type you want to configure, such as Claim, Return, or Service.
Gather the policies your team already uses, including return windows, warranty conditions, supplier requirements, and escalation criteria.
Choose an owner who will review and maintain the rules.
Create a rule
Go to Manage company → AI Agent → Rules.
Select the relevant ticket or flow type.
Add the rule in clear, plain language.
Add conditions if the rule applies only to a particular product, supplier, customer, price range, or situation.
Save the rule.
Test it on representative tickets and review the resulting action plans.
Recommended rule structure
Use the following sections where relevant:
Definition
Explain what the request type covers.
A claim is a reported problem with a product that may require troubleshooting, repair, replacement, refund, or supplier review.
How we handle it
Describe the core approach and what the AI Agent should prioritize.
Evidence and clarification
List the information needed to make a decision, such as photos, videos, serial numbers, error codes, purchase dates, or a description of troubleshooting already completed.
Typical early actions
Describe the actions normally taken while the situation is still being assessed.
Typical resolution actions
Describe the outcomes that may be used after the issue is understood, such as repair, replacement, partial refund, full refund, or supplier escalation.
Constraints
Add hard limits, such as return windows, excluded categories, value thresholds, or supplier-specific requirements.
Safety override
State which conditions always require immediate human escalation.
We avoid
Describe what the AI Agent must not do, for example promising a refund before eligibility is confirmed.
Example rule
For warranty claims involving a charging fault, first confirm that the product is within warranty and request a photo or video showing the issue if none is attached. Check for known defects and similar past cases. If the documentation is complete and the relevant product rule allows replacement, recommend a replacement. Escalate cases involving overheating, smoke, or physical injury immediately. Do not ask the customer to repeat troubleshooting already documented in the ticket.
Best practices
Write rules as handling guidance, not as rigid scripts.
Use direct language and define ambiguous terms.
Keep one canonical rule for shared behavior, then add conditions for genuine exceptions.
Include both when to take an action and when not to take it.
Start with common cases and refine the rules using feedback from real action plans.
Review rules whenever policies, products, suppliers, or workflows change.
Check your result
Open several representative tickets and review the AI Agent’s plan and reasoning. Confirm that it:
Uses the correct policy.
Requests only relevant missing information.
Respects constraints and escalation conditions.
Recommends an allowed and appropriate next step.
If the result is not correct, update the relevant rule or condition and test again before increasing automation.