Skip to content
Promptzy Promptzy

AI Assistant

Promptzy includes a built-in AI assistant that generates ready-to-use prompts on demand. It’s powered by Pollinations — a free, OpenAI-compatible AI API with access to dozens of top models.

Get a Pollinations API Key

Section titled “Get a Pollinations API Key”

  1. Visit enter.pollinations.ai and sign up for a free account.

  2. Create a new key. For browser use, choose a publishable key (pk_...) — it’s rate-limited but free and safe to use client-side.

  3. Copy your key — you’ll paste it into Promptzy’s Settings next.

Configure the AI Assistant

Section titled “Configure the AI Assistant”

  1. Open Promptzy and click the Settings icon (⚙️).

  2. Scroll to AI Assistant Configuration.

  3. Paste your Pollinations API Key into the key field.

  4. Optionally select a different AI Model (default is gemini-fast — a great all-rounder).

  5. Click Save Changes.

Generating a Prompt

Section titled “Generating a Prompt”

  1. Click AI Prompt Assistant in the bottom-right corner of the screen to expand the panel.

  2. Select a prompt type: System, Task, Image, or Video.

  3. Describe what you want in the text area. For example:

    • “a friendly customer support chatbot that speaks in a warm, casual tone” (System)
    • “write a product description for handmade soy candles” (Task)
    • “cinematic sunset over a cyberpunk city skyline, neon reflections” (Image)

  4. Click Generate Prompt — the response streams in live.

  5. Copy the result to clipboard, or click Use as Prompt to save it directly to your library.

Available Models

Section titled “Available Models”

Any Pollinations text model can be used. Change the model in Settings → AI Assistant Configuration.

ModelDescription
gemini-fastGoogle Gemini 2.5 Flash Lite — default, fast & affordable
openaiGPT-5 Mini — fast and well-rounded
openai-largeGPT-5.2 — most capable OpenAI model
claudeClaude Sonnet 4.6 — nuanced, high-quality output
claude-fastClaude Haiku 4.5 — fast and smart
deepseekDeepSeek V3.2 — efficient reasoning
mistralMistral Small 3.2 — lightweight and cost-effective
qwen-coderQwen3 Coder 30B — great for code-related prompts

Customising the System Prompt

Section titled “Customising the System Prompt”

The AI assistant uses a default system prompt tuned for generating high-quality prompts. You can override it:

  1. Open Settings → AI Assistant Configuration.

  2. Uncheck Use default system prompt.

  3. Edit the prompt freely — tailor the assistant’s persona, style, and output format to your needs.

  4. Click Save Changes. Your custom prompt is stored in your browser.

  5. Click Reset to default at any time to restore the original system prompt.

How It Works (Technical)

Section titled “How It Works (Technical)”

Promptzy calls the Pollinations API directly from your browser using a streaming fetch request:

  • Endpoint: POST https://gen.pollinations.ai/v1/chat/completions
  • Protocol: OpenAI-compatible (Server-Sent Events for streaming)
  • Auth: Authorization: Bearer <your-api-key> header
  • Default params: temperature: 0.4, seed: -1 (random), stream: true

The API key is resolved in this priority order:

  1. VITE_POLLINATIONS_API_KEY environment variable (for Docker / self-hosted builds)
  2. pollinations-api-key in localStorage (set via Settings dialog)

Pollinations API Overview

Section titled “Pollinations API Overview”

Pollinations provides access to a wide range of AI models through a single, OpenAI-compatible API.

Free to Use

Publishable pk_ keys are free with rate limits — no credit card needed for personal use.

Many Models

Access GPT-5, Claude Sonnet, Gemini, DeepSeek, Mistral, and more from one API endpoint.

Streaming

Supports Server-Sent Events (SSE) for real-time streaming responses — no waiting for the full output.

OpenAI-Compatible

Use any OpenAI SDK by just changing the base URL to https://gen.pollinations.ai.

Full API documentation: gen.pollinations.ai/api/docs