cookbook · Decider 1 · Support inbox triage

Route a support email in one call

Routing a support email usually needs several separate calls, one per field, which adds latency and cost. This recipe routes the email with a single Decider 1 call that asks three typed questions: which team (choice), how urgent (score, 4 levels), and whether a refund is requested (noul).

Request

POST /v1/systemone, documented in the Decider 1 reference.

curl https://meragpt.com/v1/systemone \
  -H "Authorization: Bearer $MERAGPT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "sd-1",
  "state": {
    "channel": "email",
    "messages": [
      {
        "from": "customer",
        "text": "Hi, the export button has been greyed out since yesterday'\''s update and I have a board meeting at 3pm where I need this report. Can someone look at it now?"
      }
    ]
  },
  "questions": {
    "team": {
      "type": "choice",
      "instructions": "Which team should handle this?",
      "criteria": {
        "billing": "Charges, invoices, payments and refunds.",
        "technical": "The product is not working as expected.",
        "account": "Login, plan changes and access."
      }
    },
    "urgency": {
      "type": "score",
      "instructions": "How urgent is this?",
      "criteria": [
        "Can wait a week",
        "Within a few days",
        "Today",
        "Immediately"
      ]
    },
    "refund_requested": {
      "type": "noul",
      "instructions": "Is the customer asking for money back?"
    }
  }
}'

What came back

The real output from running this recipe, unedited.

{
  "answers": {
    "team": {
      "type": "choice",
      "choice": "technical",
      "confidence": 0.8969,
      "probabilities": {
        "billing": 0.017,
        "technical": 0.8969,
        "account": 0.0862
      }
    },
    "urgency": {
      "type": "score",
      "score": 2.1852,
      "confidence": 0.4393,
      "legend": {
        "0": "Can wait a week",
        "1": "Within a few days",
        "2": "Today",
        "3": "Immediately"
      },
      "probabilities": {
        "0": 0.0343,
        "1": 0.1854,
        "2": 0.341,
        "3": 0.4393
      }
    },
    "refund_requested": {
      "type": "noul",
      "noul": 0.0975
    }
  }
}

Python

import os, requests

API = "https://meragpt.com/v1"
HEADERS = {"Authorization": f"Bearer {os.environ['MERAGPT_API_KEY']}"}

r = requests.post(f"{API}/systemone", headers=HEADERS, json={
    "model": "sd-1",
    "state": {
        "channel": "email",
        "messages": [
            {
                "from": "customer",
                "text": "Hi, the export button has been greyed out since yesterday's update and I have a board meeting at 3pm where I need this report. Can someone look at it now?"
            }
        ]
    },
    "questions": {
        "team": {
            "type": "choice",
            "instructions": "Which team should handle this?",
            "criteria": {
                "billing": "Charges, invoices, payments and refunds.",
                "technical": "The product is not working as expected.",
                "account": "Login, plan changes and access."
            }
        },
        "urgency": {
            "type": "score",
            "instructions": "How urgent is this?",
            "criteria": [
                "Can wait a week",
                "Within a few days",
                "Today",
                "Immediately"
            ]
        },
        "refund_requested": {
            "type": "noul",
            "instructions": "Is the customer asking for money back?"
        }
    }
}).json()

answers = r["answers"]
team = answers["team"]
if team["confidence"] >= 0.8:
    print("route to", team["choice"])
else:
    print("send to the shared queue")
print("urgency levels:", answers["urgency"]["probabilities"])

TypeScript

const API = "https://meragpt.com/v1";
const headers = {
  Authorization: `Bearer ${process.env.MERAGPT_API_KEY}`,
  "Content-Type": "application/json",
};

const r = await fetch(`${API}/systemone`, {
  method: "POST",
  headers,
  body: JSON.stringify({
    "model": "sd-1",
    "state": {
      "channel": "email",
      "messages": [
        {
          "from": "customer",
          "text": "Hi, the export button has been greyed out since yesterday's update and I have a board meeting at 3pm where I need this report. Can someone look at it now?"
        }
      ]
    },
    "questions": {
      "team": {
        "type": "choice",
        "instructions": "Which team should handle this?",
        "criteria": {
          "billing": "Charges, invoices, payments and refunds.",
          "technical": "The product is not working as expected.",
          "account": "Login, plan changes and access."
        }
      },
      "urgency": {
        "type": "score",
        "instructions": "How urgent is this?",
        "criteria": [
          "Can wait a week",
          "Within a few days",
          "Today",
          "Immediately"
        ]
      },
      "refund_requested": {
        "type": "noul",
        "instructions": "Is the customer asking for money back?"
      }
    }
  }),
}).then((res) => res.json());

const team = r.answers.team;
console.log(team.confidence >= 0.8 ? `route to ${team.choice}` : "send to the shared queue");

When to trust it

Route on the team when its probability is high, as in the example where technical scored 0.90. The urgency score is a spread, so read the top two levels rather than just the argmax; here 'Immediately' was 0.44 and 'Today' was 0.34, giving an expected level of 2.19 on the 0-3 scale.

Try it without code in the playground, or see more Decider 1 recipes: act only when the model is sure, flag a phishing email, drop retrieved passages that do not help, hold a reply that promises money, pick an agent's first tool, ask many questions in one call.