Intelligent scenario rules is a beta feature that assists users in creating rules to define optimization scenarios in Keelvar using natural language descriptions called prompts.
This feature is available on the following pages in Sourcing Optimizer:
Intelligent scenario rules can currently create rules to constrain the award or winners. Adjusting bids and other rules types are not yet available. Rule relaxation parameters, if needed, must be configured manually.
Create an Intelligent scenario rule
- Navigate to Scenarios on the Evaluate menu and click + Add a scenario
- In the Intelligent Rules section, enter a prompt describing the rule you wish to create in the Intelligent rules text box
- Select Generate rule
- One or more suggested rules are returned, each of which should be inspected by clicking on the rule in the rules list
- Inspect and, if needed, amend the rule configuration. Keep the rule if it meets your requirement. Alternatively, to discard an unwanted rule, click the three dots to the right of the rule in the rules list and select Delete
Intelligent scenario rules uses AI based on Large Language Models (LLMs). Like any output from advanced LLMs, scenarios can contain mistakes. Users must review and approve all rules and configurations to ensure accuracy before using them.
Keelvar follows best practice in AI based processing. This feature can be enabled or disabled based on your organization preference. Contact your Customer Success Manager for additional information or to enable the feature.
Usage tips
- Always check the scenario for errors and omissions. Intelligent scenario rules are created by GenAI (a Large Language Model) and as such may contain mistakes or misinterpretations of the user's intent.
- Reduce the likelihood of a misinterpretation by making a detailed and unambiguous request. For example, when referencing columns, lots, bidders or groups by name, expand any acronyms in the prompt that do not also appear the columns, bidders, lots or group names; refer to lots, groups and bidders using their exact names to avoid ambiguity.
- Imagine prompts as instructions for a novice assistant: a little more context or precision added to the request can increase the chances of getting the desired outcome.
- While generating rules, the LLM has access to limited information about the specific event and bid sheet. Specifically, it knows about the current request text, the names of the bid sheet columns, the names of lots, suppliers and any groups already created. It does not know the following: which columns are bidder input; the specific contents of bid sheet columns (e.g. the names of places in a Destination column); any past requests; any other rules or scenarios that may exist.
- Scenario rules are intended to apply business rules to an optimization scenario leveraging information that is already available in the bid sheet along with the bidder and lot groups. You may need to use the respective pages in Sourcing Optimizer to create additional bid sheet columns (such as formula columns or cost calculation columns) or additional bidder and lot groups as a prerequisite to creating a given scenario rule.
Trouble Shooting
- If no rule is returned for several minutes, an error may have occurred while processing your request. In that case, you may no longer be able to use Intelligent scenario rules with the scenario. The options are to add rules to the scenario manually or begin a new scenario to resume using Intelligent scenario rules.
- If you can see the input dialog for Intelligent Scenario rules but you are not able to submit new requests, then the feature may have become blocked due to an error processing a request. The remaining options are to add rules to the scenario manually or begin a new scenario to resume using Intelligent scenario rules.
If the input dialog is not visible, the feature may not have been enabled for you or your organization. Please contact your customer success manager to enable the feature. - The feature attempts to identify any bidders, lots or groups that are referenced in your prompt. Naming differences may cause it to find the incorrect ones. To increase the likelihood of a correct match, use the exact names given to the bidders, lots or groups you wish to reference.
- If the configuration of a generated rule is not as required, it can be adjusted manually in the rule configuration page. Consider refining your prompts to help the system better interpret your requests.
The generative AI (Large Language Model) enabling Intelligent scenario rules is privately hosted by our cloud service provider and is secure. Data submitted is not used for model training and is not available to other users.
Large Language Models can have inherent biases which may influence their output if out of scope requests are made, such as arbitrating on business decisions, or moral and ethical matters. Avoid bias risk by not making such out of scope requests.