Google removes AI Overviews for certain medical queries

✨TodayOnAI’s Daily Drop

• Google removes AI Overviews for certain medical queries

• Google announces a new protocol to facilitate commerce using AI agents

• NVIDIA Unveils New AI Memory Tier to Power Next-Gen Agentic Models

• 💬 Let’s Fix This Prompt

• 🧰 Today’s AI Toolbox Pick

Google

Google removes AI Overviews for certain medical queries

🚀 TodayOnAI Insight: Following criticism over misleading health advice, Google has quietly removed certain AI Overviews related to liver function tests after a Guardian investigation. The move signals growing scrutiny over AI-generated health content—and Google's cautious response.

🔍 Key Takeaways:

• Google removed AI Overviews for queries like “what is the normal range for liver blood tests” and “liver function tests” after reports of incomplete or potentially misleading information.

• The AI responses failed to account for key variables like age, sex, ethnicity, and nationality—factors critical to interpreting lab results.

• Alternative phrasings such as “lft reference range” initially still triggered AI Overviews but now appear disabled.

• A Google spokesperson said internal clinicians reviewed the flagged responses and found many were not inaccurate and cited high-quality sources.

• Google has been expanding health-focused features in Search, including AI-powered summaries and medically tuned models.

💡 Why This Stands Out: Google's partial rollback reflects the broader tension between AI convenience and medical safety. While AI Overviews aim to simplify complex topics, even small omissions in health contexts can have serious consequences. As generative AI becomes more embedded in everyday tools, will companies develop clearer standards for risk-sensitive domains—or continue to rely on reactive fixes?

Google

Google announces a new protocol to facilitate commerce using AI agents

🚀 TodayOnAI Insight: Google has introduced the Universal Commerce Protocol (UCP)—a new open standard for AI agent-powered shopping—at NRF 2026, aiming to streamline every step of online retail across platforms like Shopify, Etsy, Wayfair, Target, and Walmart. It’s a significant push to unify AI commerce infrastructure and accelerate agentic retail experiences.

🔍 Key Takeaways:

• UCP enables AI agents to handle the full shopping journey—from discovery to post-purchase—across multiple retailers, reducing fragmentation.

• Built with partners like Shopify, Etsy, and Walmart, the protocol supports interoperability with Google’s existing frameworks (AP2, A2A, MCP).

• Google will soon integrate UCP into product listings in AI Search and Gemini apps, allowing direct checkout via Google Pay or PayPal.

• Retailers can now offer real-time AI-triggered discounts, surfacing offers during product queries in conversational AI.

• New tools in Merchant Center and branded AI agents in Search allow retailers like Lowe’s and Reebok to better support and convert customers.

💡 Why This Stands Out: AI commerce is shifting from isolated chatbot pilots to full-stack, protocol-driven ecosystems. Google’s UCP standard signals a strategic bid to define the infrastructure layer for agentic shopping—before Amazon, Meta, or OpenAI do. With direct checkout, AI-powered discounts, and embedded brand agents, the future of retail is being architected now. Will protocols like UCP become the HTTP of digital commerce?

NVIDIA

NVIDIA Unveils New AI Memory Tier to Power Next-Gen Agentic Models

🚀 TodayOnAI Insight: NVIDIA is reshaping AI infrastructure with a new memory tier designed to scale “agentic AI”—models that reason, retain memory, and operate autonomously. Its Inference Context Memory Storage (ICMS) platform tackles a major bottleneck in handling long-term memory, offering a smarter alternative to storing vast context in expensive GPU memory or slow general-purpose storage.

🔍 Key Takeaways:

• ICMS introduces a new “G3.5” memory tier: Positioned between GPU memory and shared storage, it uses Ethernet-attached flash to store ephemeral inference context at scale.

• Built for transformer workloads: Offloads Key-Value (KV) cache—essential for agentic workflows—away from high-bandwidth memory without incurring latency penalties.

• Boosts performance and efficiency: Enables up to 5x higher tokens-per-second throughput and 5x better power efficiency for long-context inference tasks.

• Hardware-backed by BlueField-4: Offloads context management from CPUs and supports petabyte-scale shared memory per pod.

• Ecosystem support is growing: Major vendors including Dell, IBM, Pure Storage, and HPE are integrating with BlueField-4, with deployments expected in H2 2026.

💡 Why This Stands Out: Agentic AI pushes beyond chatbots into autonomous agents that rely on persistent memory and long reasoning chains—requiring infrastructure that treats memory as dynamic, not static. NVIDIA’s ICMS reflects a broader industry pivot: memory hierarchy is now a strategic priority, not a backend concern. As AI agents scale, how—and where—they remember may define the limits of what they can do.

✨ See how a simple prompt upgrade can unlock better AI output.

🔹 The Original Prompt

At first glance, this prompt might seem okay. But it's too broad — and that limits the quality of AI-generated results. Let’s improve it using prompt engineering best practices.

✅ The Improved Prompt

💡 Why It's Better

• Specific audience: Targets decision-makers in mid-sized companies.

• Contextual focus: Emphasizes emerging tech and practical solutions.

• Actionable output: Requests summaries and titles to spark execution.

• Tone and style: Guides the type of content (insightful, engaging, relevant).

🛠️ Learn how to adapt this prompt for SaaS, AI tools, dev teams & more →
Read the full PromptPilot breakdown

💡 Bonus Tool: Want to generate and master prompts instantly?
👉 Try PromptPilot by TodayOnAI (Free to use)

📰 More from the AI World

• Indonesia and Malaysia block Grok over non-consensual, sexualized deepfakes

• OpenAI is reportedly asking contractors to upload real work from past jobs

• Anthropic adds Allianz to growing list of enterprise wins

• OpenAI to acquire the team behind executive coaching AI tool Convogo

🧰 Today’s AI Toolbox Pick

• 📖PagePundit (Shopping Tool): Provides tailored book suggestions for your next read.

• 👻Uizard (Web Design Tool): Designs stunning websites, mobile apps, wireframes, prototypes, interfaces, or mockups in minutes.

• ⭕️Circleback (Note-Taking Tool): Provides AI-written action items and meeting notes you can trust.