Images & AI-generated artwork

Rewards, promotions, raffles, and levels each support an image shown in the Rigaly app. Every one has a matching image endpoint:

Resource	Endpoint
Reward	`POST /api/v1/rewards/{id}/image`
Promotion	`POST /api/v1/promotions/{id}/image`
Raffle	`POST /api/v1/raffles/{id}/image`
Level (tier)	`POST /api/v1/tiers/{id}/image`

All four accept the same three input styles — pick whichever your pipeline produces (JPEG, PNG, or WebP, max 5MB):

multipart/form-data with an image file field — a normal file upload.
JSON image_base64 — raw base64 plus content_type, or a data: URI. This is what AI image APIs return, so you can generate and attach in one flow.
JSON image_url — a public https URL that Rigaly downloads server-side.

The response returns the hosted image_url. DELETE on the same path removes it.

Create a reward with an AI-generated image

The pattern: create the reward, generate the artwork with an image model, then attach it. Because the image endpoint accepts base64 directly, the model’s output goes straight to Rigaly — no file handling, no separate hosting.

import OpenAI from "openai";
const openai = new OpenAI();
const RIGALY = { Authorization: `Bearer ${process.env.RIGALY_API_KEY}` };

// 1. Create the reward
const { data: created } = await fetch("https://api.rigaly.com/api/v1/rewards", {
  method: "POST",
  headers: { ...RIGALY, "Content-Type": "application/json" },
  body: JSON.stringify({ name: "Free Cappuccino", points_cost: 1000 }),
}).then((r) => r.json());
const rewardId = created.reward.id;

// 2. Generate artwork (returns base64)
const img = await openai.images.generate({
  model: "gpt-image-1",
  prompt: "A cozy illustrated cappuccino with latte art, warm colors, flat vector style",
  size: "1024x1024",
});
const imageBase64 = img.data[0].b64_json;

// 3. Attach it — base64 goes straight to Rigaly
await fetch(`https://api.rigaly.com/api/v1/rewards/${rewardId}/image`, {
  method: "POST",
  headers: { ...RIGALY, "Content-Type": "application/json" },
  body: JSON.stringify({ image_base64: imageBase64, content_type: "image/png" }),
});

import os, requests
from openai import OpenAI

openai = OpenAI()
RIGALY = {"Authorization": f"Bearer {os.environ['RIGALY_API_KEY']}"}

# 1. Create the reward
reward = requests.post(
    "https://api.rigaly.com/api/v1/rewards",
    headers=RIGALY,
    json={"name": "Free Cappuccino", "points_cost": 1000},
).json()["data"]["reward"]

# 2. Generate artwork (returns base64)
img = openai.images.generate(
    model="gpt-image-1",
    prompt="A cozy illustrated cappuccino with latte art, warm colors, flat vector style",
    size="1024x1024",
)
image_base64 = img.data[0].b64_json

# 3. Attach it — base64 goes straight to Rigaly
requests.post(
    f"https://api.rigaly.com/api/v1/rewards/{reward['id']}/image",
    headers=RIGALY,
    json={"image_base64": image_base64, "content_type": "image/png"},
)

Already have the image hosted somewhere? Skip base64 and pass { "image_url": "https://…" } instead.

Doing it by chatting with an AI assistant

With the AI connector, the assistant has set_reward_image, set_promotion_image, set_raffle_image, and set_tier_image tools — each takes an image_url. This works best when the artwork already lives at a public URL. For example: “Create a ‘Free coffee’ reward for 1,000 points and set its image to https://mycdn.com/coffee.png.” The assistant creates the reward and attaches the image in one go.

Images an assistant generates inside the chat usually don’t have a public URL, so it can’t hand them to the connector directly — a known limitation of today’s AI chat platforms. Until that changes, the reliable paths are: attach from a public image_url by chatting, or use the base64 API flow above in a script for fully automated “generate + attach” pipelines.

Getting Started

Guides

Images & AI-generated artwork

Create a reward with an AI-generated image

Doing it by chatting with an AI assistant

​Create a reward with an AI-generated image

​Doing it by chatting with an AI assistant

Create a reward with an AI-generated image

Doing it by chatting with an AI assistant