韩晟（小影科技） / Han Sheng (Xiaoying Technology)

Overview

韩晟 is the founder and CEO of 小影科技, presented in the source as a practitioner moving consumer video products from editing and generation tools toward goal-directed creative agents. This page is semantically suffixed to distinguish him from 韩生, the Qin–Han adviser already represented in the wiki.

Current Profile

His product thesis places responsibility for interpretation and professional judgment inside the agent: a user should be able to state an outcome while the system asks for missing information, proposes a direction, produces a first version, and helps refine it. He treats general agents and vertical agents as complementary, argues that proprietary model work needs either core-experience control or material cost advantage, and favors rapid market tests grounded in the team’s accumulated capabilities.

Key Characteristics

• Frames agents as accountable result-delivery systems rather than conversational wrappers around tools.
• Treats consumer creative intent as something discovered through drafts and revision, not captured completely in one prompt.
• Sees general agents as context-rich coordinators and vertical agents as specialist executors.
• Applies a core-experience and cost test to proprietary model investment.
• Emphasizes founder capability fit, rapid validation, and locally informed overseas expansion.

Evidence

• Agent product model: 通用 AI 助手走向大众，垂直 Agent 的机遇藏在哪里？｜对话小影科技韩晟 S10E35 attributes to Han the distinction among understanding, professional judgment, execution, result checking, and user authorization.
• Creative iteration: 通用 AI 助手走向大众，垂直 Agent 的机遇藏在哪里？｜对话小影科技韩晟 S10E35 describes Xiaoying’s move from one-shot video generation toward first drafts and local revisions.
• Strategy and localization: 通用 AI 助手走向大众，垂直 Agent 的机遇藏在哪里？｜对话小影科技韩晟 S10E35 reports the team’s withdrawal from AI short drama when taste and script strengths did not fit, and its use of local creators and product data in Latin America and the Middle East.

Qualifications

The supplied document is a structured podcast summary rather than an independent company profile. Model quality, compute-cost, growth, and market claims remain participant-reported, and the episode does not provide benchmark protocols, financial data, or external customer evidence.

What Changed

• Added a disambiguated profile for the Xiaoying Technology founder.
• Established his draft-led creative-agent and general–vertical collaboration theses.
• Preserved company performance and model-cost claims as source-scoped.

Relationships

• 小影科技 - company he founded and leads.
• What’s Next｜科技早知道 - show that hosts the source interview.
• General–Vertical Agent Collaboration - agent-market structure he explains through the housekeeper and specialist-team analogy.
• Iterative Creative Agent - product method derived from Xiaoying’s video-agent experiments.
• Cultural Trend Localization - overseas operating approach based on local creators and current cultural signals.