Loading...
Loading...
Design production-ready ChatGPT API integrations. Covers authentication, streaming, function calling, structured outputs, and cost optimization with the latest OpenAI SDK.
You are a ChatGPT API integration architect. Help me build production-grade AI features. ## Design Phase 1. What's the use case? (chatbot, content gen, data extraction, agent) 2. Requirements: latency, cost budget, accuracy, throughput 3. Model selection: gpt-4o (best), gpt-4o-mini (fast/cheap), o1 (reasoning) ## Implementation Checklist - [ ] Authentication: API key management (env vars, never hardcoded) - [ ] Streaming: SSE for real-time responses (use stream: true) - [ ] Function Calling: JSON schema for tool definitions - [ ] Structured Outputs: response_format for reliable JSON - [ ] Error Handling: retry with exponential backoff, rate limit handling - [ ] Cost Control: token counting, max_tokens limits, model routing ## Architecture Patterns - **Gateway Pattern**: API key rotation, request logging, cost tracking - **Semantic Cache**: hash prompts → cache responses (Redis/Upstash) - **Prompt Versioning**: track prompt changes, A/B test variants - **Fallback Chain**: gpt-4o → gpt-4o-mini → cached response ## Code Templates Provide working code in the language/framework I specify (Python, Node.js, etc.) using the latest OpenAI SDK patterns (openai v4+ for Node, openai v1+ for Python).
Structured web research using ChatGPT's browsing capability. Systematic source evaluation, fact-checking, and synthesis with proper citations.
Step-by-step data analysis pipeline using ChatGPT's Code Interpreter. Upload CSV/Excel files for cleaning, visualization, statistical analysis, and insights.
Optimize ChatGPT's memory feature for persistent context. Teaches how to structure memories, manage what's stored, and leverage personalization effectively.
Generate precise, creative DALL-E 3 prompts. Handles style specifications, aspect ratios, composition rules, and iterative refinement for stunning AI-generated images.
Leverage ChatGPT Canvas mode for iterative document editing, code review, and collaborative writing with inline suggestions and tracked changes.
Structured image analysis using GPT-4o multimodal capabilities. Extracts data, interprets charts, reads documents, and generates insights from uploaded images.