AI Startup Stories: Hippocratic AI - Large Language Model (LLM) for healthcare

Hippocratic AI, a healthcare AI startup, raised a $141 million Series B round at a $1.64 billion valuation, led by Kleiner Perkins. This brings its total funding to $278 million. The company develops generative AI agents for non-diagnostic, patient-facing tasks like pre-op procedures, remote patient monitoring, and chronic care management to address the global healthcare staffing shortage.

Since launching in May 2023, Hippocratic AI has rapidly expanded, signing contracts with 23 health systems and insurers, going live with 16, and completing over 200,000 patient interactions. The company emphasizes AI safety, achieving parity with human clinicians through rigorous multi-phase testing with thousands of licensed nurses and physicians.

Hippocratic AI introduced a healthcare AI agent app store, allowing clinicians to create AI agents for various medical tasks. Clinicians can design agents without programming skills and earn revenue from their use. The company aims to expand into new verticals, including pharma, and expand internationally.

The startup's AI agents have been used for patient check-ins during natural disasters, chronic disease monitoring, and maternal mental health. Investors praise its focus on safety and strong commercial traction, positioning it as a leader in AI-driven healthcare staffing solutions.

Hippocratic AI banks $141M series B, hits 'unicorn' status
Hippocratic AI raises $141M for creating patient-facing AI agents

AI Resources: Hugging Face AI Agent Course

AI is evolving fast, and AI Agents are at the forefront of this transformation. Whether you're a developer, researcher, or just an AI enthusiast, understanding how AI Agents work and how to build them is becoming an essential skill.

That's why I'm excited about Hugging Face’s free AI Agents Course—a hands-on learning journey that takes you from beginner to expert in using and building AI Agents.

Why AI Agents Matter

AI Agents go beyond simple prompts and responses. They can reason, plan, and act autonomously, making them the foundation for next-gen applications like:
✅ Autonomous research assistants
✅ Smart customer support bots
✅ AI-driven workflow automation
✅ Agents that integrate with real-world tools and APIs

This course doesn’t just teach theory—you’ll get hands-on experience with leading AI agent frameworks like LangGraph, LlamaIndex, and smolagents while building and testing your own AI Agents.

What You’ll Gain

📖 Deep understanding of AI Agent design and functionality
🧑‍💻 Practical experience with real-world libraries
💾 The ability to share and deploy agents on the Hugging Face Hub
🏆 The chance to compete in challenges and benchmark your skills
🎓 A free certificate for completing the course

Link to course

Open Source Alternative to Manus AI Agent

OWL by CAMEL-AI.org is an open-source alternative to Manus AI Agent, designed for multi-agent collaboration and real-world task automation. It ranks #1 on the GAIA Benchmark among open-source AI agent frameworks with a score of 58.18.

Key Features:

• Multi-Agent Collaboration: AI user agents break down tasks, assistants create execution plans, and tool agents interface with external APIs.

• Advanced Automation: Supports autonomous research, coding, web browsing, multimodal processing, browser automation, and document parsing.

• Flexible Model Integration: Works with Claude Sonnet 3.7, GPT-4o, DeepSeek, Gemini, and Ollama.

• Easy Installation & Setup: Installable via conda, uv, Docker, with TOML-based configuration.

• Free & Open-Source: No paywalls, invites, or subscriptions.

Use Cases:

• Information Retrieval & Processing

• Web Automation

• Document Analysis

• AI-Assisted Coding

• Multimodal AI Tasks (images, video, audio)

• Prototyping AI Systems & Research in Multi-Agent AI

With a web-based UI, built-in toolkits, and strong community engagement, OWL is positioned as a cutting-edge AI agent framework for robust automation and research in multi-agent AI systems.

Link to the repo

JFrog and Hugging Face Team to Improve AI Security and Transparency

JFrog has partnered with Hugging Face to improve security scanning for machine learning (ML) models hosted on the Hugging Face Hub. This integration aims to reduce false positives by 96% while detecting real threats that traditional scanners often miss. JFrog’s advanced security tools, including JFrog Xray, scan models for risks such as serialization attacks, backdoors, and known vulnerabilities.

False positives in existing ML model scanners cause unnecessary alerts, while false negatives allow malicious models to go undetected. JFrog’s approach, using malicious code decompilation and deep data flow analysis, enhances security by accurately identifying harmful models. Their analysis has already identified 25 zero-day malicious models on Hugging Face that other scanners overlooked.

The collaboration addresses growing AI security concerns, as ML models become critical for businesses but remain vulnerable to cyber threats. JFrog’s system not only flags potential threats but also explains why a model is deemed safe or dangerous. Additionally, their secure proxy, JFrog Curation, blocks downloads of high-risk models to protect users.

This partnership highlights the need for stronger AI security, with JFrog and Hugging Face leading efforts to secure the ML supply chain.

Link to article

Additional details

OpenAI calls for DeepSeek and Other Chinese Model Ban

OpenAI has labeled Chinese AI lab DeepSeek as "state-controlled" in a proposal to the Trump administration, recommending a ban on its models in "Tier One" countries. The company cites security risks from Chinese data processing laws that could compel user data sharing with the government. While DeepSeek originated from hedge fund High-Flyer rather than a government institution, recent meetings between its founder and President Xi Jinping suggest growing government interest. This proposal would escalate the ongoing technological tensions between the US and China in the AI field.

Link to article