I am using copilot studio flow to extract some entities using LLM(*Run a prompt* in json mode).
The text in prediction output is proper json format with the field names as expected(as given in the json response in the prompt), but the structuredOutput tends to have some additional suffix to some of the keys.
{
"@odata.type": "#Microsoft.Dynamics.CRM.expando",
"<ENT>fa3141a378143db37921086cc12e251c": "",
"<ENT>": "",
"<ENT>_002e63aea1947726491770f09c383cab0": "",
"<ENT>_af8f5241e3acd73f60de122bca5a6b4b": "",
"<ENT>@odata.type": "#Collection(Microsoft.Dynamics.CRM.crmbaseentity)",
"<ENT>": [
{
"@odata.type": "#Microsoft.Dynamics.CRM.expando",
"<ENT>": "",
"<ENT>": {
"@odata.type": "#Microsoft.Dynamics.CRM.expando",
"<ENT>_002a0e82e2a18101defdf6a7f4fb0376267": "",
"<ENT>_0092667e014bfa4aeaad20d33565ef8403": "",
"<ENT>_0076724a15e386177369b8f6e3929ef528": "",
"<ENT>": "",
"<ENT>": "",
"<ENT>": "",
"<ENT>": ""
}
},
]
}
What could be the reason?