Domo Hands On: Building AI Marketing Agent Error

I was following the AI Academy: Episode 2 - Your Marketing Agent Has Entered The Chat to the T. However, when testing the AI Agent task, it did everything from reading the datasets up till crafting the email except it didn't want to output the result to emailBody & emailSubject. Can anyone spot what went wrong?
{
"outputMessages": [
{
"content": [
{
"type": "TEXT",
"text": "Use the following datasets to create a tailored and persuasive marketing campaign that is relevant to "Bargain Hunters"\n\nCustomer Segment dataset: 23bd7900-793d-4577-963e-2656a0e2a845\n\nDiscount code dataset: 45f60159-31fd-4883-a270-ff285f8e9ab1\n\nProduct list dataset: 27768405-05e8-40c3-8812-5621a693f736\n\nActivities and Events dataset: 69ae0c5e-8bb8-4aa2-803f-1ef80fa2acdd"
}
],
"role": "USER"
…
{
"content": [
{
"type": "TEXT",
"text": "Based on the upcoming events and activities, the Tech Recycling Week on 2025-05-20 seems most relevant to include for the "Bargain Hunters" segment. Here's how I would incorporate it:\n\n{\n"emailSubject": "Upgrade Your Tech & Save Big with Our Recycling Deals!",\n"emailBody": "Dear Valued Customer,\n\nGet ready to save big during our upcoming Tech Recycling Week from May 20th-27th! As a bargain-hunting pro, you know a good deal when you see one. \n\nThis is your chance to trade in your old electronics and receive up to $150 in store credit to use towards your next purchase. Whether it's a new TV, laptop, or must-have accessory, we've got you covered with unbeatable prices.\n\nBut that's not all! In addition to the lucrative trade-in offers, we're sweetening the deal by giving you 30% off clearance items when you use the code CLEARANCE30 at checkout. \n\nDon't miss out on this opportunity to save big and upgrade your tech in an eco-friendly way. Stop by our store during Tech Recycling Week and let our experts guide you through the trade-in process.\n\nSee you soon!\n\n[Store Name]\n[Address]"\n}\n\nThe key points are:\n\n1) Highlighting the up to $150 store credit for trading in old electronics, appealing to their bargain hunting nature.\n\n2) Reminding them about the 30% off clearance items with the CLEARANCE30 code to really maximize savings.\n\n3) Promoting it as an opportunity to upgrade their tech affordably while being environmentally conscious.\n\n4) Encouraging them to visit the store for guidance on the trade-in process.\n\nThis combines the recycling event, clearance discount code, and their preference for deals into one compelling promotion for the "Bargain Hunters"."
}
],
"role": "ASSISTANT"
},
{
"content": [
{
"type": "TEXT",
"text": "Now will you help me format this into my desired output response? Please use all context of this solution and the inputs and outputs that best match my output schema"
}
],
"role": "USER"
},
{
"content": [
{
"type": "TOOL_USE_RESULT",
"toolCallId": null,
"content": [
{
"type": "TEXT",
"text": "null"
}
]
}
],
"role": "USER"
}
],
"result": null
}
Answers
-
Hi there- I'm in the same boat but I think even less far along. I posted here (see below). Can you help?
0 -
Try editing your prompt to this and see if you have any better luck:
(the "Custom" "Input" under "Parameters" in the CONFIGURATION tab of the AI Agent Task) to this:
The customer segment called "
CustomerSegment
" has been selected for a marketing campaign.Please use the following datasets to create a tailored and persuasive marketing campaign that is relevant to this specific customer segment.
Customer Segments dataset =
customerSegments
Discount Code dataset =
discountCodes
Product List dataset =
productList
Activities & Events dataset =
events
0 -
@BryantCafferty - Thanks for the updates !! I tried playing around with the settings, but the result got worse as it errored out even earlier. Frustrated, I switched off everything, and all that remained was as below. Refreshed my screen and tested again.
Similar to your tests, Bargain Hunters error out all the time. "Home Tech Builders" seems to work better for me. But overall, the success rate stands at 66.6% :(
0 -
OK, yeah. That's helpful to hear you've also seen inconsistency with success or failure based on the selected segment name. I take that to mean (since sometimes it works!) that the fragility is in the AI portion where it can crap out depending on the prompt and the ask and the datasets, etc.
0
Categories
- All Categories
- 2K Product Ideas
- 2K Ideas Exchange
- 1.6K Connect
- 1.3K Connectors
- 311 Workbench
- 6 Cloud Amplifier
- 9 Federated
- 3.8K Transform
- 657 Datasets
- 115 SQL DataFlows
- 2.2K Magic ETL
- 815 Beast Mode
- 3.3K Visualize
- 2.5K Charting
- 81 App Studio
- 45 Variables
- 775 Automate
- 190 Apps
- 481 APIs & Domo Developer
- 81 Workflows
- 23 Code Engine
- 40 AI and Machine Learning
- 20 AI Chat
- 1 AI Playground
- 1 AI Projects and Models
- 18 Jupyter Workspaces
- 410 Distribute
- 120 Domo Everywhere
- 280 Scheduled Reports
- 10 Software Integrations
- 144 Manage
- 140 Governance & Security
- 8 Domo Community Gallery
- 48 Product Releases
- 12 Domo University
- 5.4K Community Forums
- 41 Getting Started
- 31 Community Member Introductions
- 114 Community Announcements
- 4.8K Archive