AI tools

E-Smartpoints integrates multiple AI services to help you create professional material without being a design expert. The tools run on two Cloudflare Workers:

• example-WORKER.JS — image generation, editing, video, text and templates
• fal-WORKER.JS — optimized image and video generation with Fal.ai

Integrated services

The platform uses the following services depending on the task:

Servicio	Uso	Modelo(s)	Worker
🔵 OpenAI	Generación de imágenes publicitarias	gpt-image-1.5	example-WORKER.JS
🟢 Gemini	Generación de imágenes, video y texto	gemini-3.1-flash-image-preview (imagen), veo-3.1-generate-preview (video), gemini-2.5-flash (texto)	example-WORKER.JS
🟤 Fal.ai	Generación de imágenes y video	Recraft v3 (imagen), Pixverse v6 (video), LTX 2.3 Fast (video)	fal-WORKER.JS
🔧 Poof	Eliminación de fondo	api.poof.bg	example-WORKER.JS
🎨 Orshot	Plantillas de marketing	Orshot Studio API	example-WORKER.JS
🤖 Puter	Generación de logos	gpt-image-1 vía Puter AI	Logo Generator Worker

Available tools

🎨 Background Removal (Poof)

Remove the background from any product image:

1. Upload an image
2. Poof service processes the image
3. Receive the product without a background (transparent PNG)
4. Download the result

Service used: Poof (api.poof.bg)

✨ Add object (Object AI)

Add elements to an existing image:

1. Upload the base image
2. Describe the object to be added
3. Specify position and size
4. AI integrates the object naturally

Models used: OpenAI (gpt-image-1.5) or Gemini (gemini-3.1-flash-image-preview) with auto rotation

📢 Ad Generator

Create a full ad:

Inputs

• 🏢 Business name
• 📝 Description
• 🎯 Target audience
• 🎨 Ringtone (professional, casual, fun, premium)
• 🏷️ Industry/category
• 💡 Additional details
• 🎁 Special coupon (optional)
• 🌈 Brand colors (optional)

Outputs

The system generates:

• 📝 Caption in 3 paragraphs
• 🏷️ Relevant Hashtags
• 🖼️ Image of the advertisement
• 📐 Format optimized for each platform

Models used: OpenAI (gpt-image-1.5) and Gemini (gemini-3.1-flash-image-preview) with rotation. Fal.ai (Recraft v3) for generation with reference image.

🎬 Video generator

Convert a static image into a short advertising video:

1. Upload an image
2. A video prompt is built with the trade data
3. AI generates the video with camera animations and text overlay
4. Brand elements are integrated (logo, QR, name)
5. Download the MP4

Used models:

• Gemini (veo-3.1-generate-preview) — 15-second advertising video with German voiceover, text overlay and branded endcard
• Fal.ai (pixverse/v6/image-to-video or ltx-2.3/image-to-video/fast) — 5-second UGC-style video optimized for social media

🖼️ Templates (Orshot)

Use Orshot studio templates to generate professional designs:

1. Select a template from the Orshot gallery
2. Upload your brand assets (logo, colors)
3. Orshot renders the final design
4. Download or publish

Service used: Orshot API

🤖 Logo generation (Puter)

During business registration you can generate a logo with AI:

1. Complete brand name, business type, style and description
2. Puter AI generates the logo using OpenAI's gpt-image-1 model
3. The image is automatically uploaded as your logo

Limit: 2 generations per month. Counter stored in IndexedDB.

AI Endpoints

example-WORKER.JS

Endpoint	Método	Función
/ai/image/generate	POST	Generación de imágenes (OpenAI/Gemini con rotación)
/ai/image/generate-ad	POST	Generación de anuncios con Gemini
/ai/image/edit	POST	Edición de imágenes
/ai/image/remove-background	POST	Eliminación de fondo (Poof)
/ai/image/remove-object	POST	Eliminación de objetos
/ai/video/generate-from-image	POST	Generación de video desde imagen (Gemini Veo)
/ai/text/gemini	POST	Generación de texto (Gemini)
/orshot/templates	GET	Listar plantillas de Orshot
/orshot/render	POST	Renderizar plantilla de Orshot

fal-WORKER.JS

Endpoint	Método	Función
/ai/image/generate	POST	Generación de imágenes (Recraft v3)
/ai/video/generate-from-image	POST	Generación de video (Pixverse v6 o LTX 2.3)
/ai/task/status/{taskId}	GET	Estado de tarea asíncrona
/ai/image/upload	POST	Subir imagen temporal
/ai/image/temp/{imageId}	GET	Obtener imagen temporal
/ai/video/temp/{videoId}	GET	Obtener video temporal

Infrastructure

• Cloudflare Workers: run AI models at the edge
• Upstash Redis: temporary storage of tasks and assets (fal-WORKER.JS)
• Cloudinary: storage of published feeds (images and videos)
• Cloudflare Turnstile: anti-bot verification for AI endpoints
• IndexedDB: control of usage limits per user (logos)

Typical workflow

1. Identify what you need to create
2. Choose the right tool
3. Provide the minimum inputs
4. Review the results
5. Iterate if necessary (adjust prompts, regenerate)
6. Post to the app feed or download

Best practices

📝 For texts

• Be specific in your descriptions
• Includes the target audience
• Mention unique benefits
• Avoid unnecessary technical jargon

🖼️ For images

• Use good quality base images (>1000px)
• Avoid multiple products in one photo
• Good natural lighting
• Clean background (the service removes it anyway)

💡 For prompts

• Describes the desired result, not the process
• Use specific adjectives (modern, minimalist, vibrant)
• Specify the emotional tone (professional, close, fun)
• Iterates if the first result is not perfect

Limitations

• 📊 Cost: each use consumes credits. See your plan in Subscriptions
• ⏱️ Time: generation can take 5-30 seconds (video: up to 2 minutes)
• 🔄 Iterations: maximum 3 regenerations per image
• 🚫 Content: violent, sexual or copyright violating content is not allowed

Next steps

• Return to marketing module →
• Know your subscription plan →