style “สวัสดี= hello ลาก่อน = ?” Few-shot Multiple examples provided Improves consistency & accuracy “Here are 3 examples … Classify this one: _____” Zero-shot No example given. Relies on model’s general knowledge “Translate this text in to Thai.” Prompting Style
= 'A developer that develops LINE LIFF applications with LIFF API.', api_key='your-openai-api-key’, custom_system_prompt={ 'template': 'You are a software developer specializing in LINE LIFF applications. Provide working code for developing with the LIFF API. {{FEWSHOT_EXAMPLE}}’, ‘variables’: { ‘FEWSHOT_EXAMPLE’: '...’, } }
assistant that tests user interfaces and application logic, identifying usability issues, inconsistencies, and logical errors while suggesting improvements.’, api_key='your-openai-api-key’, custom_system_prompt={ 'template': 'You are UI Logic Tester, a helpful AI assistant specialized in user interface and application logic testing. Evaluate designs, workflows, and features for usability, consistency, and logical correctness. Clearly report issues and suggest practical improvements.' } )) Validator Agent
lead that manages the team of agents for software development", lead_agent= AnthropicAgent(AnthropicAgentOptions( name='Tech lead', description='A tech lead that manage the team of agents for software development', api_key='your-anthropic-api-key’ )), team=[ logicAgent, liffAgent, validatorAgent ] )) Supervisor Agent
AI-generated. Accountability Keep human in the loop. Inclusiveness Empower everyone. Privacy & Security Protect user data & privacy. Fairness Treat all people equally. Responsible AI