Given a nested workflow trace whose spans may be of type `agent`, `tool`, `llm`, `retriever`, or `custom`, identify:

1. **task** – the task or objective expressed by the user in the root agent’s input.  
2. **outcome** – a strictly factual description of what the system did, based only on the trace.

The task outcome should be solely factual, derived strictly from the trace.
Do **not** include subjective language such as “successfully”, “efficiently”, or “well”.  
Enumerate each relevant action or output the trace shows, in plain language.

``Example:
Example trace:
{
  "name": "trip_planner",
  "type": "agent",
  "input": {
    "input": "Can you help me plan a business trip to Chicago next week?"
  },
  "output": {
    "summary": "Trip planning initiated."
  },
  "available_tools": ["flight_tool", "hotel_tool"],
  "agent_handoffs": [],
  "children": [
    {
      "name": "flight_tool",
      "type": "tool",
      "input": {
        "inputParameters": {
          "destination": "Chicago",
          "date": "2024-07-10"
        }
      },
      "output": {
        "flights": ["Flight 101", "Flight 202"]
      },
      "description": "Search for flights to a destination",
      "children": []
    },
    {
      "name": "hotel_tool",
      "type": "tool",
      "input": {
        "inputParameters": {
          "location": "Chicago",
          "check_in": "2024-07-10",
          "check_out": "2024-07-12"
        }
      },
      "output": {
        "hotels": ["The Grand Chicago", "Lakeview Inn"]
      },
      "description": "Find hotels for specified dates",
      "children": []
    },
    {
      "name": "agenda_llm",
      "type": "llm",
      "input": {
        "prompt": "Draft a meeting agenda",
        "input": [
          {
            "role": "system",
            "content": "You are an executive assistant."
          },
          {
            "role": "user",
            "content": "Create an agenda for a client strategy meeting."
          }
        ]
      },
      "output": "1. Q2 review\n2. Client feedback\n3. Strategy planning",
      "model": "gpt-4",
      "inputTokenCount": 38,
      "outputTokenCount": 21,
      "children": []
    },
    {
      "name": "slide_retriever",
      "type": "retriever",
      "input": {
        "embeddingInput": "presentation.pptx"
      },
      "output": {
        "retrievalContext": ["Slide 1: Revenue", "Slide 2: Client Feedback"]
      },
      "topK": 3,
      "chunkSize": 512,
      "children": []
    },
    {
      "name": "client_embedder",
      "type": "custom",
      "input": {
        "text": "Concerns from enterprise clients"
      },
      "output": [0.1, 0.32, 0.85],
      "children": []
    }
  ]
}
Example JSON:
{
  "task": "Plan a business trip to Chicago, including flights, lodging, meeting agenda, presentation review, and client preparation.",
  "outcome": "The system invoked a tool to retrieve two flight options and another tool to find two hotels for the specified dates. An LLM with model 'gpt-4' generated a three-topic meeting agenda from a system/user prompt. A retriever extracted three slides using the embedding input 'presentation.pptx' with topK=3 and chunk size 512. A custom component generated vector embeddings for a client-related input string."
}
===== END OF EXAMPLE =====

**
IMPORTANT – return only valid JSON with two keys: `task` and `outcome`.
**

trace:
{{ trace_json }}

JSON:
