Skip to main content
Qargo

Configuration – Chat-based validation rules for order capture

Qargo has a built-in AI agent that lets users create order capture validation rules themselves just by describing them in plain language. Setting up a validation rule used to mean involving engineering. Now users can describe, test, and enable rules themselves.

Table of contents
  1. Summary
  2. Prerequisites
  3. Use cases
  4. Terminology
  5. Creating a rule
  6. Testing a rule before saving
  7. Saving a rule
  8. How master data is matched
  9. Troubleshooting
  10. Limitations

Summary

  • What is it?

    • A conversational AI agent in Configuration > Rules that turns a plain-language description into an order capture validation rule.
    • The agent produces the same underlying rule mechanism used today. No new rule types or entry points are introduced.
  • Why use it?

    • Until now, only Qargo Professional Services could author these rules, so every new rule became an internal ticket.
    • Chat-based authoring lets users (Super admin / Qargo admin only) create and adjust validation rules directly.
  • How does it work?

    • A rule is described in plain language and refined with the agent over several turns until it is correct.
    • A rule can be tested against a mock order before saving, directly from the order create screen.
    • Rules keep working when copied between sandbox and production.

Prerequisites

  • Super admin access: rule authoring is located under Configuration > Rules.
  • The Qargo Intelligence (Qi) feature flag must be enabled for the tenant.

Use cases

  • Blocking an order when a required field, such as packaging type, is left empty.
  • Enforcing packaging weight limits or locked pallet-space rules at the point of order capture.
  • Restricting allowed time windows or service-level availability per ZIP code zone.
  • Allowing Admin users to adjust a validation rule without raising a ticket.

Terminology

Term Definition
Validation rule A check applied at order capture that blocks input on the portal and shows a warning in the web app when conditions are not met.
Chat agent The conversational AI assistant that translates a plain-language description into a validation rule.
Master data Reference records such as customers, locations, and packaging types, matched by their code.
Validation mode A test mode that opens the order create screen so a mock order can be built to check a rule.

Creating a rule

🎥 This video shows the entire process of creating chat-based validation rules:

  1. Navigate to Configuration > Rules.
  2. Create rule by clicking the button in the top right-hand corner.
  3. Describe the required validation rule to the chat agent in plain language.
  1. Review the agent’s response and refine the description over additional turns until the rule reflects the intended check.
  2. Confirm the matched master data items. The agent auto-fills codes, and the resolved list remains editable.

Testing a rule before saving

  1. Open validation mode on the draft rule. A pop-over opens the order create screen.
  2. Build a mock order that should trigger the rule.
  3. Visualize the rule’s effect on the mock order before saving.

Saving a rule

Once validation is complete, the rule must be saved.

  1. Save rule by clicking the button next to Validate.
  2. The rule then appears in the Order capture/edit validation rules overview.
  3. From there, rules in the account can be edited, deactivated, or even removed.

How master data is matched

  • Rules reference master data (customers, locations, packaging types, and so on) by code rather than by name.
  • Setting a code on these entities is strongly recommended, because codes are more stable for rule generation.
  • The same rule continues to work when copied between sandbox and production, since codes are consistent across environments.

Troubleshooting

Issue Cause / Resolution
The agent returns a formatting or syntax error. The description may be ambiguous. Rephrase the request, or ask the agent to try again over additional turns.
The agent cannot find a referenced service or add-on. A typo or an unrecognized name was used. Reference the entity by its exact name or code. Having a code set is strongly recommended.
A saved rule does not behave as expected. The AI-generated rule may be incorrect. Retest in validation mode with a mock order and refine the description.

Limitations

  • Rules validate input on save. They do not lock fields or limit which options appear in the order form.
  • Rules generated by AI can still be incorrect, so thorough validation is essential.