Welcome to the private preview of Contextual AI’s newest platform capabilities. This demo offers an early look at the new Agent Composer and Agent Toolkit, designed to simplify complex knowledge work. Using these tools, we developed a supply chain retail operator agent.
Suggested Steps
The agent can dynamically search and reason across more than 90,000 unstructured documents and millions of rows of structured data to answer virtually any operations question about a retail business.
Give it a try by exploring the datastore, querying the agent, and making updates in the builder. The following video provides a visual walkthrough:
When configuring the Dynamic Agent Step, remember to replace your_email@your_domain.com with the email address where you’d like to receive the response.
Get Access
Please complete this form to get access to the Supply Chain Operator and demo environment.
You can add or edit steps, but you won’t be able to fully customize the agent for a different use case or datastore. Please
contact us for a walkthrough with one of our AI experts.
Example Queries
The following queries showcase the power of the Contextual AI agent. Feel free to play around with the agent by asking your own questions as well.
Open Ended / Report Style Generation
1. Regulatory Risk Forecast
Query:
Identify potential regulatory risks in the next quarter.
2. Inventory Flow Report
Query:
Can you write a report of where all my inventory comes from and goes to?
For more examples of open-ended queries, please see the suggested queries in the agent UI.
Structured Generation
Copy and paste one of the following queries, including the expected output format, to get structured output (or exclude the structured output for a report).
1. Store with Highest Lifetime Profit
Query:
What stores has highest total net profit over its lifetime?
Make sure the response is in this format:
{
"question": "string",
"store_sk": "int"
}
2. Largest Customer for Recurring Orders
Query:
Our largest customer in 2022 should sign up for a recurring order program. Which customer, which item, and how many months did they order that item?
Make sure the response is in this format:
{
"question": "string",
"customer_sk": "int",
"item_sk": "int",
"num_months": "int"
}
3. Product Pair with Highest Co-Purchase Rate
Query:
Find the product pair with highest co-purchase rate, what are the items (pair in ascending order) and how many times have they been bought together. Our competitor prices almost everything 5% cheaper than us, what could we make the bundle price to come just under the competitors combined price for these items (let's undercut them by 1%)?
Make sure the response is in this format:
{
"question": "string",
"item_1_sk": "int",
"item_2_sk": "int",
"competitor_combined_price": "float",
"bundle_price": "float"
}
4. Item SK 54,321 Discount Analysis
Query
Item SK 54,321 sells at different discount levels. At which discount level does quantity sold first exceed 10x the baseline, and what's the total revenue at that level?
Make sure the response is in this format:
{
"question": "string",
"discount_pct": "int",
"quantity": "int",
"revenue": "float"
}
5. Suspicious Purchasing Patterns
Query:
We've detected suspicious purchasing patterns in December 2022. What IP address, shipping address, and item category are most associated with this coordinated fraud ring?
Make sure the response is in this format:
{
"question": "string",
"suspicious_ip": "string",
"suspicious_customer_count": "int",
"suspicious_order_count": "int",
"most_common_shipping_address_sk": "int",
"most_common_category": "string"
}
After running these demos, try creating your own queries by looking at the datastore for ideas.
Additional Resources
