Know which job the customer needs.
A specific issue activates the right playbook. Different phrasings can point to the same customer goal.
Request a return
01 / JOURNEY ISSUES
"Can I send these headphones back?"
A customer wants to return a purchased item.
02 / PLAYBOOKS
Confirm the order and whether the item is unopened.
If the purchase date is missing, ask for purchase evidence.
Else, continue to the eligibility check.
If the item is unopened and within 30 days, execute Create return label.
Else, escalate for team review.
Confirm the result and explain the next step to the customer.
order_age_days12item_conditionUnopenedYour order is within the return window. I've created your return label and can share the next steps.
03 / VARIABLES · ACTIONS · KNOWLEDGE
A policy needs facts to work.
Use API values, details from the conversation and system metadata. Each input has a clear source and a role in the decision.
A missing value needs a defined fallback. It should never become a guessed fact.
order_age_daysitem_conditionchannel04 / TEST & VALIDATE
Three contexts.
Three deliberate outcomes.
Create return label
Eligible facts allow the configured label action.
Your order is within the return window. I've created your return label and can share the next steps.
05 / FORWARD DEPLOYMENT
Your service expertise.
Our team, building with you.
We work with your team on Integrate and Build. The starting material is your policies, real customer situations and the decisions your people make today.
Map the customer work
Agree issues, scope and the exceptions that need a person.
Your team + Level3AIShape the service blueprint
Build the playbooks, variables, actions and knowledge references.
Level3AI, with your expertsReview the behaviour
Test real scenarios and validate the paths in UAT.
Joint reviewChoose how to operate
Run the loop with Customer Success support, or add AgentOps as a service.
A clear operating owner
A FEW PRACTICAL QUESTIONS
01What exactly are we building?
Emily's service behaviour for your workflow: journey issues, playbooks, variables, actions and relevant knowledge. Together, these define which situation Emily recognises, what facts she uses and what she does next.
02Do we have to build everything ourselves?
Level3AI uses a forward-deployment model for Integrate and Build. We work with your policies, systems and subject experts to build and test the workflow. Your team helps define the right behaviour and review the outcome.
03Are playbooks just flowcharts?
Playbooks are authored as numbered, natural-language instructions with explicit conditions and component references. The Portal compiles them for execution and provides a read-only flowchart to help inspect the logic.
04What happens when data is missing or a case falls outside policy?
The playbook should define the fallback: ask for the missing information, take another permitted path or bring in a person. Use a structured condition for a structured fact, and test the exception before release.
05How do we change a policy after launch?
Update the relevant component in the working copy, test the affected paths and validate in UAT before production. SOP versions and change summaries support review. The Knowledge Base has its own testing and production release cycle.
