He Tao

He Tao is the core person behind Deerflow in 对话 MiniMax 闫俊杰：M3、10X 计划、10T 模型、和智能的终局. He uses the project to discuss open-source AI workflow building, codebase governance, long-term engineering responsibility, and the spread of AI into family and personal-care contexts.

Key Views

• Vibe coding should not be mistaken for engineering; durable software requires long-term maintainability.
• One-off SWE benchmarks do not fully measure whether agents can sustain and evolve a real codebase.
• Clean Code, Google best practices, Amazon best practices, and similar engineering standards can be turned into AI Skills or agent context.
• Developers remain responsible for code committed under their accounts, even if an agent generated it.
• AI can help ordinary people and families by organizing private context such as medical records, medications, and questions for doctors.

Connections

• Deerflow — project and community context.
• AI Coding Verification — engineering and maintenance concern.
• AI Skills — possible packaging for durable software practices.
• Human Judgment Under AI — personal responsibility and situated judgment.
• Financial AI Agents — adjacent theme of AI as domain-specific companion and guide.