Minara, an AI fintech company, made a significant appearance at ACM SIGKDD 2026 in Jeju, South Korea...
The AMW Read
Minara is a new entrant in AI fintech, updating the player map; the benchmark highlights industry-wide gaps in models, but the event is a product/benchmark reveal with limited novelty. Significance is segment-level as it impacts AI finance and evaluation practices.
Minara, an AI fintech company, made a significant appearance at ACM SIGKDD 2026 in Jeju, South Korea, unveiling its vision for a "Self-Evolving Financial OS" during an invited talk and presenting the DMind Benchmark paper, which was accepted into the conference's Datasets & Benchmarks Track. The company also appeared as an official sponsor alongside Alibaba and Meta. The DMind Benchmark, developed with Zhejiang University and Nanyang Technological University, evaluates LLMs across nine professional subdomains related to Web3, including blockchain infrastructure and smart contract security, with over 3,500 expert-reviewed questions. It assessed 31 major models, including GPT, Claude, and Gemini, and found that even the most advanced models exhibit significant capability gaps in multi-step reasoning and complex rule-constrained scenarios. The benchmark, released on Hugging Face in April 2025, has accumulated over 13,000 downloads. During the conference, Minara gained over 1,000 new registered members, reflecting strong interest from developers.
This event matters in the AI market as it signals a trend toward domain-specific evaluation and self-improving AI agents, particularly in high-stakes financial decision-making. Minara's focus on a "self-evolving" financial operating system, which emphasizes continuous learning, autonomous decision-making, risk control, and human-AI collaboration, addresses a critical need for reliable AI in complex professional environments. The DMind Benchmark's finding that frontier models still struggle with multi-step, rule-based reasoning highlights the necessity for robust evaluation frameworks and targeted improvements. This positions Minara not only as a fintech player but also as a contributor to AI governance and trust, aligning with broader industry pushes for verifiable AI capabilities.
For builders and investors, the implication is that the next wave of AI value will come from vertical-specific evaluation and continuous learning systems, especially where accuracy and compliance are non-negotiable. Startups that can combine domain expertise with AI agent engineering, and demonstrate measurable progress through rigorous benchmarks, may gain a competitive edge. Investors should watch for companies building "self-evolving" loops around proprietary data and feedback, as they could dominate niches like financial operations. Minara's success at KDD also validates the strategy of early academic collaboration and open benchmarking to establish credibility in the AI ecosystem.