Quickstart

Your first agent run in five minutes

Step 01

Configure the agent

Build the agent in the panel first — its instructions, which model it runs on, which tools it can call and what it knows. None of this is passed at call time; the agent already holds it.

GTWY panel · New agent

NameSupport agent

InstructionsCheck the order, apply the refund policy

Modelclaude-sonnet-4-6 · fallback gpt-4o

Toolsorders.lookup · stripe.refund

Knowledgerefund-policy.pdf

Test it in the panel before you publish. It is a much faster loop than testing through the API.

Step 02

Publish and get your auth key

Publishing is what makes the agent callable. You get an agent ID and an auth key — the key is shown once, so store it somewhere your server can read and nowhere else.

GTWY panel · Published

Agent IDagt_4127

Auth keygtwy_live_9f4c…a17e

StatusLive

Keep the auth key on your server. Never ship it to a browser — that is what embed tokens are for.

Step 03

Make the first call

Name the agent, send an input. Everything the agent needs it already has — the snippet below fills in a real prompt with two variables, so you can see the variables block populate.

First call

curl --location 'https://api.gtwy.ai/api/v2/model/chat/completion' \
  --header 'pauthkey: YOUR_GENERATED_PAUTHKEY' \
  --header 'Content-Type: application/json' \
  --data '{
  "user": "YOUR_USER_QUESTION",
  "agent_id": "agt_4127",
  "thread_id": "YOUR_THREAD_ID",
  "response_type": "text",
  "variables": {
    "order_id": "YOUR_ORDER_ID_VALUE",
    "customer_name": "YOUR_CUSTOMER_NAME_VALUE"
  }
}'

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: "YOUR_GENERATED_PAUTHKEY",
  baseURL: "https://api.gtwy.ai/api/v2/model/openai",
});

const response = await client.responses.create({
  model: "YOUR_MODEL_NAME",
  agent_id: "agt_4127",
  input: "YOUR_USER_QUESTION",
  variables: {
    "order_id": "YOUR_ORDER_ID_VALUE",
    "customer_name": "YOUR_CUSTOMER_NAME_VALUE",
  },
});

console.log(response.output_text);

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_GENERATED_PAUTHKEY",
    base_url="https://api.gtwy.ai/api/v2/model/openai",
)

response = client.responses.create(
    model="YOUR_MODEL_NAME",
    input="YOUR_USER_QUESTION",
    extra_body={
        "agent_id": "agt_4127",
        "variables": {
            "order_id": "YOUR_ORDER_ID_VALUE",
            "customer_name": "YOUR_CUSTOMER_NAME_VALUE",
        },
    },
)

print(response.output_text)

using System;
using System.Threading.Tasks;
using OpenAI;

class Program
{
    static async Task Main()
    {
        var client = new OpenAIClient(
            apiKey: "YOUR_GENERATED_PAUTHKEY",
            options: new OpenAIClientOptions
            {
                Endpoint = new Uri("https://api.gtwy.ai/api/v2/model/openai")
            }
        );

        var response = await client.Responses.CreateAsync(new ResponseCreateRequest
        {
            Model = "YOUR_MODEL_NAME",
            Input = "YOUR_USER_QUESTION",
            AdditionalProperties =
            {
                ["agent_id"] = BinaryData.FromString('"agt_4127"'),
                ["variables"] = BinaryData.FromObjectAsJson(new
                {
                ["order_id"] = BinaryData.FromString('"YOUR_ORDER_ID_VALUE"'),
                ["customer_name"] = BinaryData.FromString('"YOUR_CUSTOMER_NAME_VALUE"')
                })
            }
        });

        Console.WriteLine($"[ASSISTANT]: {response.OutputText()}");
    }
}

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.responses.Response;
import com.openai.models.responses.ResponseCreateParams;

public class Main {
    public static void main(String[] args) {
        OpenAIClient client = OpenAIOkHttpClient.builder()
            .apiKey("YOUR_GENERATED_PAUTHKEY")
            .baseUrl("https://api.gtwy.ai/api/v2/model/openai")
            .build();

        ResponseCreateParams params = ResponseCreateParams.builder()
            .model("YOUR_MODEL_NAME")
            .input("YOUR_USER_QUESTION")
            .putAdditionalBodyProperty("agent_id", "agt_4127")
            .putAdditionalBodyProperty("order_id", "YOUR_ORDER_ID_VALUE")
            .putAdditionalBodyProperty("customer_name", "YOUR_CUSTOMER_NAME_VALUE")
            .build();

        Response response = client.responses().create(params);
        System.out.println(response.outputText());
    }
}

package main

import (
	"context"
	"fmt"

	"github.com/openai/openai-go/v3"
	"github.com/openai/openai-go/v3/option"
	"github.com/openai/openai-go/v3/responses"
)

func main() {
	client := openai.NewClient(
	option.WithAPIKey("YOUR_GENERATED_PAUTHKEY"),
	option.WithBaseURL("https://api.gtwy.ai/api/v2/model/openai"),
	)

	resp, err := client.Responses.New(context.TODO(), openai.ResponseNewParams{
		Model: "YOUR_MODEL_NAME",
		Input: responses.ResponseNewParamsInputUnion{OfString: openai.String("YOUR_USER_QUESTION")},
	},
		option.WithJSONSet("agent_id", "agt_4127"),
		option.WithJSONSet("variables", map[string]string{
		"order_id": "YOUR_ORDER_ID_VALUE",
		"customer_name": "YOUR_CUSTOMER_NAME_VALUE",
		}),
	)
	if err != nil {
		panic(err.Error())
	}

	fmt.Println(resp.OutputText())
}

Response

{
  "success": true,
  "response": {
    "data": {
      "id": "chatcmpl-d7a6874d-a82f-4cb5-8a40-1c899722c64f",
      "content": "Response from the AI assistant",
      "model": "your-model-name",
      "role": "assistant",
      "tools_data": {},
      "fallback": false,
      "finish_reason": "completed",
      "message_id": "abdd920a-ec69-11f0-b14a-928ade59a1ee"
    },
    "usage": {
      "total_tokens": 500,
      "input_tokens": 300,
      "output_tokens": 200,
      "cached_tokens": 0,
      "cache_read_input_tokens": 0,
      "cache_creation_input_tokens": 0,
      "reasoning_tokens": 0,
      "cost": 0.0025
    }
  }
}

{
  "id": "resp_d7a6874d-a82f-4cb5-8a40-1c899722c64f",
  "object": "response",
  "created_at": 1771660421,
  "status": "completed",
  "model": "your-model-name",
  "output": [
    {
      "id": "rs_de18a639a6144712a27a33c3fbe5ab17",
      "type": "reasoning",
      "summary": []
    },
    {
      "id": "msg_b051e3c65248486e97044615a7d3e779",
      "type": "message",
      "status": "completed",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "Response from the AI assistant"
        }
      ]
    }
  ],
  "user": "YOUR_USER_QUESTION",
  "output_text": "Response from the AI assistant",
  "finish_reason": "completed",
  "usage": {
    "input_tokens": 300,
    "input_tokens_details": {
      "cached_tokens": 0
    },
    "output_tokens": 200,
    "output_tokens_details": {
      "reasoning_tokens": 64
    },
    "total_tokens": 500
  }
}

{
  "id": "resp_d7a6874d-a82f-4cb5-8a40-1c899722c64f",
  "object": "response",
  "created_at": 1771660421,
  "status": "completed",
  "model": "your-model-name",
  "output": [
    {
      "id": "rs_de18a639a6144712a27a33c3fbe5ab17",
      "type": "reasoning",
      "summary": []
    },
    {
      "id": "msg_b051e3c65248486e97044615a7d3e779",
      "type": "message",
      "status": "completed",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "Response from the AI assistant"
        }
      ]
    }
  ],
  "user": "YOUR_USER_QUESTION",
  "output_text": "Response from the AI assistant",
  "finish_reason": "completed",
  "usage": {
    "input_tokens": 300,
    "input_tokens_details": {
      "cached_tokens": 0
    },
    "output_tokens": 200,
    "output_tokens_details": {
      "reasoning_tokens": 64
    },
    "total_tokens": 500
  }
}

{
  "id": "resp_d7a6874d-a82f-4cb5-8a40-1c899722c64f",
  "object": "response",
  "created_at": 1771660421,
  "status": "completed",
  "model": "your-model-name",
  "output": [
    {
      "id": "rs_de18a639a6144712a27a33c3fbe5ab17",
      "type": "reasoning",
      "summary": []
    },
    {
      "id": "msg_b051e3c65248486e97044615a7d3e779",
      "type": "message",
      "status": "completed",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "Response from the AI assistant"
        }
      ]
    }
  ],
  "user": "YOUR_USER_QUESTION",
  "output_text": "Response from the AI assistant",
  "finish_reason": "completed",
  "usage": {
    "input_tokens": 300,
    "input_tokens_details": {
      "cached_tokens": 0
    },
    "output_tokens": 200,
    "output_tokens_details": {
      "reasoning_tokens": 64
    },
    "total_tokens": 500
  }
}

{
  "id": "resp_d7a6874d-a82f-4cb5-8a40-1c899722c64f",
  "object": "response",
  "created_at": 1771660421,
  "status": "completed",
  "model": "your-model-name",
  "output": [
    {
      "id": "rs_de18a639a6144712a27a33c3fbe5ab17",
      "type": "reasoning",
      "summary": []
    },
    {
      "id": "msg_b051e3c65248486e97044615a7d3e779",
      "type": "message",
      "status": "completed",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "Response from the AI assistant"
        }
      ]
    }
  ],
  "user": "YOUR_USER_QUESTION",
  "output_text": "Response from the AI assistant",
  "finish_reason": "completed",
  "usage": {
    "input_tokens": 300,
    "input_tokens_details": {
      "cached_tokens": 0
    },
    "output_tokens": 200,
    "output_tokens_details": {
      "reasoning_tokens": 64
    },
    "total_tokens": 500
  }
}

{
  "id": "resp_d7a6874d-a82f-4cb5-8a40-1c899722c64f",
  "object": "response",
  "created_at": 1771660421,
  "status": "completed",
  "model": "your-model-name",
  "output": [
    {
      "id": "rs_de18a639a6144712a27a33c3fbe5ab17",
      "type": "reasoning",
      "summary": []
    },
    {
      "id": "msg_b051e3c65248486e97044615a7d3e779",
      "type": "message",
      "status": "completed",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "Response from the AI assistant"
        }
      ]
    }
  ],
  "user": "YOUR_USER_QUESTION",
  "output_text": "Response from the AI assistant",
  "finish_reason": "completed",
  "usage": {
    "input_tokens": 300,
    "input_tokens_details": {
      "cached_tokens": 0
    },
    "output_tokens": 200,
    "output_tokens_details": {
      "reasoning_tokens": 64
    },
    "total_tokens": 500
  }
}

id: Identifies this run in logs, traces and the usage API.
content / output_text: What the agent produced. A string unless you asked for structured output.
role: Always assistant on success — useful when parsing generically.
model: Which model actually served it — useful when fallback fired.
usage.cost: The run fee, deducted from your wallet as it executed.

What just happened

That last call was one run.

You sent one input. The agent you configured called a model, ran its tool, retrieved from its knowledge, checked the output and returned it. Five things, one run, one charge.

01Request inYour input, your auth key, the agent ID
02Model callThe agent decides what to do
03Tool callorders.lookup, with retries
04KnowledgeRetrieval over your documents
05GuardrailsChecked before it leaves
06Response outWith its trace and its cost

Your first agent run in five minutes

Configure the agent

Publish and get your auth key

Make the first call

That last call was one run.

Where to go from here.

Models

Full docs

Builders wall

Get your API key.