Kando AI
Category: AI Agents
Kando AI builds a self-evolving AI decision-making workbench — the 'Cursor for decision-making' — that learns from real-world financial and research decisions by capturing user feedback and post-hoc outcomes. Kando AI was founded in 2026. The company is led by 毛书翰 (Mao Shuhan). Based in Shenzhen, China. Team size: 1-10. Total funding raised: $5.0M. Latest round: Seed. Key investors include 星连资本 (Xinglian Capital), 力合金控 (Leaguer Financial Holdings / Lihe Financial Holdings).
- Founded
- 2026
- Headquarters
- Shenzhen, China
- Team size
- 1-10
- Total funding
- $5.0M
Value proposition
Kando AI is the 'Cursor for decision-making' — an AI workbench that doesn't just process information but learns from how users actually make decisions, capturing adoption/rejection patterns and post-hoc outcomes to continuously improve its recommendations.
Products and solutions
Kando AI Decision Workbench — an AI-powered workbench for high-value decision-making scenarios (finance, scientific research). Features include: cognitive memory system (beyond traditional RAG), decision trajectory tracking, post-hoc outcome analysis, recursive learning loop (judgment→action→feedback→update), personalized skill scheduling and model adaptation.
Unique value
Unlike traditional AI tools that optimize output quality (summarization, search, writing), Kando AI optimizes decision adoption rate and post-hoc reliability — building a recursive closed loop where each decision outcome improves the system's next recommendation.
Target customer
Professional users making high-frequency, high-value decisions: financial investors (secondary market), scientific researchers, and other knowledge workers in non-standardized, cognition-intensive domains.
Industries served
Financial services (investment research, trading), Scientific research, Future expansion into any high-value, non-standardized decision domain
Technology advantage
Self-developed infrastructure layer (training, feedback, and memory pipelines are fully transparent and auditable); Cognitive memory system that goes beyond traditional RAG to capture user understanding patterns, preferences, and judgment revision history; 'Generalized post-training' approach where the product itself is treated as a continuous training process; Team combines top AI research (Peking University CS PhD, Tencent AI Lab) with domain expertise in finance and embodied intelligence.
How they differentiate
Kando AI differentiates by focusing on decision adoption rate and post-hoc reliability rather than output quality. Its memory system captures not just what users viewed but how they understood problems, their preferences, and how they revised judgments. The system forms a recursive closed loop where each decision outcome improves the next recommendation — turning the product itself into a continuous post-training process.
Main competitors
Traditional financial research platforms (TongHuaShun, Wind), AI research assistants and knowledge management tools, General-purpose AI copilots (ChatGPT, Claude) used for research
Major milestones
June 2026: Company founded, July 2026: Completed seed round financing (tens of millions RMB), July 2026: Product launched in internal testing (内测)
Market positioning
Early-stage startup positioning itself as the 'Cursor for decision-making' — analogous to how Cursor revolutionized AI coding by achieving 60-70% code adoption rates, Kando aims to achieve similar adoption rates in financial and research decision-making.
Geographic focus
China (domestic market initially)
About 毛书翰 (Mao Shuhan)
Co-founder/CEO Mao Shuhan: Tsinghua University BS; University of Hong Kong MS in Finance; former senior role at top-tier Chinese investment bank; co-founded an embodied intelligence startup; Forbes 30 Under 30 2025
