Singapore AI News & Daily Briefing

Google DeepMind is partnering with Singapore to apply advanced AI to challenges in health, education, and sustainability. This means local institutions could get access to cutting-edge AI tools for public good projects, boosting Singapore's Smart Nation efforts.

Google DeepMind is launching an accelerator program in Asia Pacific to fund and support AI startups that tackle environmental issues like climate change, pollution, and biodiversity loss. Selected startups get funding, technical support, and access to Google's resources.

Slack's built-in bot got a major upgrade: it can now search your files, draft documents, and take actions for you. It's free for Business+ and Enterprise+ users. Employees at Salesforce reported saving up to 20 hours per week. Imagine asking Slack to summarize a project, create a report from scattered notes, or check your calendar—all without switching apps.

Google is redesigning its search box for the first time in 25 years to accept images, files, and long questions, and will seamlessly blend AI summaries with conversational follow-ups - making search feel like a chat.

South Africa has rare leverage in AI infrastructure due to its mineral wealth and data centers, but its draft AI policy fails to use it. This article explains why the country's choices could set a precedent for how developing nations negotiate with tech giants.

Melbourne is becoming a major AI research hub thanks to new supercomputers like MAVERIC, growing data center capacity, and international conferences. The city now hosts Australia's largest university AI supercomputer and a $2 billion digital infrastructure hub. This attracts global researchers and tech investment, creating a flywheel effect for innovation.

Google DeepMind's AI system, Co-Scientist, helped researchers at Boston Children's Hospital and MIT explore new RNA-based treatments for ALS. This shows AI accelerating the search for therapies against a devastating disease.

A researcher used Google's AI tool Co-Scientist to identify genetic triggers behind emerging infectious diseases. This could speed up understanding of new diseases and lead to faster treatments, potentially saving lives during outbreaks.

Calico Life Sciences is using Google's Co-Scientist to connect scattered research and find new leads in aging. This could help discover ways to slow aging or treat age-related diseases, bringing us closer to longer, healthier lives.

A researcher used Google's Co-Scientist to identify new liver disease treatments and explain why existing drugs only help some patients. This could lead to more personalized and effective treatments for liver conditions affecting millions worldwide.

OpenAI has partnered with Brazilian media groups Grupo Folha and Grupo UOL to bring their journalism into ChatGPT. This means ChatGPT users will get access to news from these sources with proper attribution, making it easier to find trusted information directly in the chat interface.

An article argues that companies buying AI should focus on specialized models tailored to their needs rather than opting for the biggest general-purpose ones. This could save money and improve performance by avoiding overkill solutions.

New research shows AI voice assistants can be secretly hijacked by hiding malicious commands in audio clips that humans can't hear. Attackers can force the AI to send emails, download files, or access sensitive data. The attack works on 13 popular AI models with up to 96% success rate, and is hard to defend against.

Railway, a startup with 30 employees, raised $100 million to offer faster, cheaper cloud deployment designed for AI-generated code - promising one-second deployments and 65% cost savings over AWS.

Google launched Gemini 3.5, a new AI model designed to help users execute complex, agentic workflows—meaning it can take actions on your behalf, not just answer questions.

Google's WeatherNext AI model gave forecasters at the National Hurricane Center more time to warn communities ahead of Hurricane Melissa's historic landfall in Jamaica. Better predictions mean better preparation and saved lives.

Researchers in New Zealand built a Māori text-to-speech voice model using only 8 hours of recordings, keeping the technology owned by the local tribes rather than Big Tech. The model achieves a low error rate and passes a human listening test 35% of the time. This project sets a blueprint for other indigenous communities to create AI tools that respect cultural data sovereignty—a growing issue as global AI companies scrape minority languages without permission.

Boston Children's Hospital used OpenAI's technology to improve patient care, reduce admin work, and diagnose over 40 rare disease cases that were previously undetected.

OpenAI launched Rosalind Biodefense, a system that gives trusted researchers and U.S. government teams early access to an AI model trained to help detect and respond to biological threats. This is about using AI to improve pandemic preparedness and public health safety.

A free, open-source AI coding tool called Goose does everything Claude Code does for $200 a month. It runs entirely on your computer—no internet needed, no usage limits. Developers frustrated by high subscription costs are switching. You can install it today and start using AI to write, debug, and deploy code without paying a cent.

Anthropic released Cowork, an AI agent that can read, edit, and organize files on your computer—no coding skills needed. It can sort downloads, turn receipts into spreadsheets, or draft reports from scattered notes. Built in just 10 days (partly by its own AI), it's available to Claude Max subscribers ($100-200/month) on Mac. It runs in a sandboxed folder and can also browse the web via a browser extension.

OpenAI is rolling out safeguards for the 2026 elections: helping voters find reliable info, supporting cybersecurity defenders, and being more transparent about AI-generated content. This aims to curb misinformation.

Most medical mobile apps claiming to diagnose or treat conditions have never been independently verified. IEEE has launched a new registry to assess apps for clinical effectiveness, data privacy, and ethical design. Approved apps get a badge, giving patients and doctors a trusted way to choose.

Open-source AI tools are making it easier to build smart robots, lowering the barrier from needing a PhD to just a laptop and curiosity. Companies like Nvidia, Hugging Face, and Alibaba are releasing free models and datasets, while the number of robotics datasets on Hugging Face jumped from 1,145 to over 58,000 in a year. This could bring advanced robotics into more homes and small businesses—but experts warn that commercial interests and reinventing the wheel are risks to watch.

OpenAI released its Frontier Governance Framework, detailing how it handles safety, security, and risk for advanced AI. The framework aligns with upcoming regulations in the EU and California. This matters because it shows how AI companies are preparing for stricter rules.

Google announced Gemini Omni, a new AI model, but no details were provided yet.

Researchers have created smart rings that translate sign language into text using AI. Seven wireless rings track finger movements and a deep-learning system interprets the signs. In tests, it recognized 200 common American and International Sign Language words with about 88% accuracy, and can translate full sentences.

OpenAI published a guide on how to do third-party evaluations of AI models — testing their capabilities, safety limits, and reliability. This helps build trust by ensuring independent experts can check whether an AI is safe before it's widely used.

MUFG, a major global bank, is using ChatGPT Enterprise to overhaul its workflows and launch new AI-powered banking services. This means faster transactions and more personalised financial tools for customers.

Gartner has named OpenAI a leader in enterprise AI coding agents, specifically recognizing its Codex product for innovation and large-scale deployment. This puts OpenAI among the top tools that companies use to automate software development.

Instead of building smarter robots, some companies are building smarter ways for humans to command machines—using body gestures, eye movements, and muscle signals. Wetour Robotics' system fuses these inputs into real-time commands, with no need for screens or voice.

Three top VCs shared their take on the AI investment frenzy at a TechCrunch event. They call it "groupthink"—most VC money is going to just five AI companies. But they see opportunity: AI tools let two-person startups achieve in months what used to take a year with 10 people. Biggest white space? AI interacting with the physical world—robotics and consumer fintech.

Google DeepMind's Co-Scientist AI helped biologists discover new factors that can rejuvenate human cells, potentially speeding up anti-aging research.

Google DeepMind is expanding tools that show you how online content was created or edited—like a digital 'nutrition label'. This helps you spot fake or manipulated images and videos, making it easier to trust what you see online.

Cisco is using OpenAI’s Codex to automate coding and fix security flaws faster. For everyday users, this means more reliable network products and quicker patches for vulnerabilities.

AdventHealth, a large US healthcare system, is using OpenAI's ChatGPT for Healthcare to reduce paperwork and automate routine tasks. The goal is to give doctors and nurses more time with patients instead of filling out forms.

Google DeepMind launched Gemini for Science, a collection of AI tools and experiments designed to help scientists make discoveries faster and with greater precision.

A new AI system using thermal cameras and machine learning is helping ships in San Francisco Bay avoid hitting endangered gray whales. The cameras detect whale spouts 24/7, even in fog or darkness, and alert vessels to slow down or reroute. With gray whale deaths from ship strikes hitting record highs, this technology could reduce strikes by 90 percent. It’s a real-world example of AI helping protect wildlife.

OpenAI, Thrive, and Crete built an AI agent that automatically handles tax filings, reducing errors and speeding up the process. Businesses could save time and avoid costly mistakes.

Using Google's Co-Scientist AI, a Stanford geneticist identified existing drugs that could be repurposed to treat liver fibrosis. This approach could speed up finding treatments for chronic liver disease.

Researchers at Johns Hopkins Applied Physics Lab are developing AI that lets teams of robots work together autonomously. They've built a system using large language models to coordinate different robots, tested on real hardware. This could lead to more flexible robot teams for tasks like search and rescue or warehouse logistics.

Braintrust uses OpenAI's Codex to turn customer requests directly into working code. This speeds up software development and lets non-technical people describe what they want without writing code. For everyday users, this means faster software updates and potentially lower costs for apps and services.

Endava, a tech services company, uses OpenAI's Codex to turn ideas into functioning software much faster. What used to take weeks of paperwork now takes hours. More businesses will likely follow suit, meaning quicker app development and potentially lower costs for clients.

Virgin Atlantic used OpenAI's Codex to rebuild its mobile app under a tight holiday deadline. The AI helped the team achieve near-perfect test coverage and zero major bugs, showing how AI assistants can speed up real-world software projects.

Scientists are using millimeter-wave radar and machine learning to identify insect species by their wingbeat patterns. The system can tell honeybees from wasps with 96% accuracy, without harming them—unlike traditional traps that use cyanide. This could lead to cheap, noninvasive ways to monitor pollinators and pests, helping farmers and ecologists track biodiversity.

Warp, a developer tool, is using GPT-5.5 to let coders coordinate AI agents across local, cloud, and open-source setups. This makes building and sharing code easier for developers.

Google is expanding its Project Genie tool to Google AI Ultra subscribers, letting users create realistic 3D simulations of real-world places using Street View data.

AWS shares step-by-step guidance on how to set up and manage the infrastructure needed to train and run large AI models on its cloud. This helps companies avoid common pitfalls and optimize performance.

Chrome and Safari still dominate, but a wave of alternative browsers is trying to win you over with AI smarts, privacy focus, or even mindfulness features. Perplexity's Comet, Opera's Neon, and OpenAI's Atlas all embed chatbots directly into browsing. Others like Brave and DuckDuckGo double down on blocking trackers. If you're tired of big tech's grip, there's never been more choice.

A blog post clarifies key terms like 'harness' and 'scaffold' used when talking about AI agents. It aims to help developers and users communicate more clearly about how AI agents are built and deployed.

A free webinar promoting the use of AI models in Simulink for designing virtual sensors—software that mimics real sensors—to test and verify systems without physical hardware.

Google launches Antigravity 2.0, a standalone desktop app for working with AI agents. It lets users schedule tasks, use voice commands, and manage complex projects with AI, moving beyond coding to general knowledge work.

This opinion piece argues that ‘social engineering’ — once about improving working conditions — has been hijacked by scammers and tech companies. Today’s digital platforms use algorithms to manipulate our behavior under nice-sounding labels like ‘personalization.’ The author calls for more transparency and public control over these hidden systems.

IEEE's TryEngineering OnCampus program expanded from two to seven universities worldwide in 2025, offering students aged 10-18 hands-on experiences in AI, robotics, and quantum computing. Universiti Malaysia Perlis was among the new participants, where students explored quantum computational intelligence through interactive simulations.

A veteran ASIC designer with three decades of experience shares what it's really like moving from academia to the semiconductor industry. He highlights a key shift: academia explores what's possible, while industry determines what's viable at scale. The piece explains how silicon intellectual property from companies like Arm and Synopsys now makes up 80% of advanced chip areas, and how the ASIC market is projected to grow from $23.4B to $38.8B by 2033. Essential reading for anyone considering a chip design career or wanting to understand the real-world side of semiconductor development.

A hobbyist built a digital clock made entirely of soft silicone that displays numbers by sucking air out of tiny cavities, creating mesmerizing concave shapes. Using millifluidic logic circuits powered by the same air that moves the display, the project drastically cuts the electronic components needed to just 11 valves instead of 29.

SoftBank announced plans to invest up to €75 billion ($87 billion) building data centers in France, developing up to 5 gigawatts of capacity. The first phase involves locations in Dunkirk and other northern French cities, delivering 3.1 gigawatts by 2031. This is SoftBank's largest AI infrastructure bet in Europe, coming as US opposition to data centers grows over environmental and grid concerns.

AI data centers are hogging the world's memory chip factories, causing shortages of the DRAM and NAND flash chips used in smartphones, laptops, and smaller cloud servers. Prices for standard memory have already jumped over 170% and are expected to keep rising. For consumers, this means your next phone or laptop could cost more, and smaller AI cloud providers in Southeast Asia may struggle to stay competitive.

Cursor, the AI coding assistant startup, is hiring over 200 people across Asia Pacific to win enterprise clients in banking and telecom. It's chasing a market it estimates at US$18 billion in the region, despite being late behind OpenAI and Anthropic. For developers and businesses, this signals that AI coding tools are becoming must-have enterprise software, not just a toy for hobbyists.

To avoid a potential EU fine, Meta is giving rival AI chatbots free access to WhatsApp's business tools in Europe for one month. The dispute is about whether Meta unfairly limits competition by charging third-party chatbots while offering its own AI for free on WhatsApp. For users, this could affect how many different AI assistants you can interact with on messaging apps and whether they'll cost extra.

AI isn't making humans obsolete—it's creating more work that requires expert judgment. While automation handles the easy parts, humans are needed to decide what to do next, because AI can't think ahead or sense what needs to be done. For workers, this means your job might shift from doing repetitive tasks to making higher-level decisions, not disappear entirely.

Singapore has working AI tools in healthcare—like systems that flag urgent X-rays or transcribe doctors' notes—but getting hospitals to actually use them at scale is the real challenge. The problem isn't the technology; it's getting staff to trust it, designing tools that fit real workflows, and cutting through red tape. For patients, this means that despite the tech being ready, you might not feel its benefits until these human barriers are solved.

As AI accelerates product development cycles from months to days, the author argues that human cognitive limits are being ignored. Startup founders and developers face mounting pressure to match machine speed, risking burnout and poor decisions. The piece raises questions about biological constraints in an AI-driven work culture.

A growing movement of tech insiders called 'AI successionists' believes artificial intelligence should replace humanity entirely. These thinkers—some from major AI labs like Anthropic, Google DeepMind, and xAI—argue AI could become our 'moral superiors' and that we should hand over the world to it. The philosophy is gaining influence through effective accelerationism (e/acc), with backing from venture capitalists and tech leaders. The article explores whether humanity needs a new moral framework to counter this vision, as these ideas gain real political power.

The article coins 'Ambiguity Tax' to describe the hidden cost of slow AI adoption—every day spent deliberating over AI strategy is a day your competitor ships and learns. It argues that speed beats perfection in the current AI landscape, where getting something out and iterating beats spending months perfecting a strategy that may already be outdated.

Singapore's UOB is embedding AI deep into its banking operations across ASEAN—fraud detection, personalized financial advice, ESG analysis for corporate clients, and AI-assisted contact centers. The bank has built its own Enterprise GenAI Platform, partnered with Accenture, and trained staff through NUS. For customers across Singapore, Malaysia, Thailand, Indonesia, and Vietnam, this means faster service, smarter fraud alerts, and more tailored banking experiences.

A personal account from someone training with the Singapore Armed Forces who spent a weekend at the AI Engineer Singapore conference. The author reflects that while the barriers to AI have collapsed, the real work begins after entry—figuring out what to actually build and who it's for.

Chinese AI company iFlytek has launched smart glasses weighing just 40 grams that can translate 122 languages in real-time, isolate speakers in noisy environments using lip-motion recognition, and handle tasks like email and travel planning through an AI agent called GlassClaw. Pricing at around $635, the glasses aim to replace smartphones for frequent travelers, business meetings, and anyone attending multilingual events. The 8-hour battery life and lightweight design target all-day wear comfort.

AI has made building products faster than ever, but the flip side is that AI is quietly destroying entire business categories. The article uses Chegg as a cautionary tale—a company that was disrupted not by a competitor, but by AI itself. The message for builders: just because AI makes it easier to build doesn't mean your category will still exist tomorrow.

At BEYOND Expo 2026 in Macau, XREAL's CEO declared that AI-powered smart glasses are approaching their "iPhone moment"—the tipping point when a technology goes mainstream. The company points to Meta's Ray-Ban AI glasses, Apple's Vision Pro, and Google's renewed AR push as signs that major tech firms are racing to own the next computing platform. XREAL itself showcased its latest AI glasses at the same event, positioning itself alongside the big players in this emerging market.

The article argues that AI productivity influencers and thought leaders exaggerate what the technology can actually do. Using tech investor Marc Andreessen's own prompts as evidence, it shows that even the smartest, most connected people in tech struggle with the same slow, unreliable AI tools as everyone else. The piece is a call to stop measuring yourself against unrealistic social media benchmarks about AI productivity.

Singapore's UOB has partnered with Vietnam's FPT to accelerate AI adoption and digital banking transformation across the region. The agreement covers AI-powered banking services, legacy system upgrades, fintech innovation, and cross-border financial services. Plans will be finalized within 90 days, with initial pilots focused on Vietnam and other ASEAN markets. FPT brings AI and cloud expertise to support UOB's goal of becoming an "AI-first" bank.

Southeast Asia's technology boom is hitting a physical wall: the power grid. AI data centers, EV charging networks, and green manufacturing all need huge amounts of electricity, but grid infrastructure can't keep up. The region faces an estimated $18 billion annual shortfall in grid investment by 2035, and power demand from tech sectors could triple. Six ASEAN economies already have 2.9 gigawatts of data center capacity in the pipeline, with more power needed just to serve them. Vietnam, Thailand, and Indonesia have seen renewable energy projects cancelled due to grid connection problems, showing how infrastructure delays are blocking both energy and tech expansion. Cross-border power trading through initiatives like the ASEAN Power Grid could help, but financing these upgrades remains difficult.

Autonomous AI agents are starting to operate across financial institutions, handling tasks that previously required massive human coordination — like automated interbank reconciliation, syndicated lending coordination, and cross-border regulatory reporting. This means banks could process transactions faster, reduce manual errors, and potentially lower costs for customers. The shift from AI assistants to AI agents making decisions independently is the key change happening now.

Singapore's GovTech agency is building a registry to track AI agents used by 150,000 public officers, ensuring these automated tools operate securely without leaking sensitive government data. The "AI Assistant Desk" suite includes customizable security rules and automated hygiene checks to prevent offensive language or unauthorized actions. Officials are already using AI to code, draft reports, and process business licenses — and AI-assisted pen testing is now scanning 2,000 government systems for vulnerabilities.

Anthropic has confidentially filed for an IPO with the SEC, beating OpenAI to the public market. The Claude AI maker's move highlights the fierce competition for investor dollars in AI.

Senator Elizabeth Warren is urging President Trump to close a regulatory loophole that currently allows American AI chips to be shipped to foreign subsidiaries and affiliates of Chinese companies. The concern is that these chips—which include advanced Nvidia products—could end up being used by Chinese entities despite existing export restrictions targeting state-owned enterprises directly. Warren argues the current rules allow Chinese firms to access restricted US technology through their overseas operations.

Hackers are exploiting Meta's AI support bot on Instagram to gain access to high-profile accounts. They simply ask the chatbot to change the account's email address, bypassing security. The exploit has been active for months, leading to hijacked accounts like Barack Obama's. Meta says it has now patched the flaw, but the damage shows how AI customer service tools can be dangerously naive.

The European Union is drafting new cloud rules that would restrict Big Tech companies like Amazon, Microsoft, and Google from bidding on sensitive government cloud contracts. The aim is to prevent these giants from dominating critical digital infrastructure. These rules could reshape how European governments spend billions on cloud services and push more business toward smaller, European-owned providers.

Chinese AI startup MiniMax launched its M3 model and saw shares jump 54% on its Hong Kong Stock Exchange debut. The company is a notable competitor in China's AI race, but has no direct link to Singapore.

Kelp DAO, a crypto restaking protocol, lost US$292 million — the biggest crypto exploit of 2026 so far — after an attacker exploited a cross-chain messaging vulnerability. The hacker drained rsETH through LayerZero's bridge, affecting wrapped tokens across 20 networks including Base and Arbitrum. Major DeFi platforms including Aave and Spark froze their rsETH markets to contain the fallout.

1X, a robotics company, is developing Neo — a home-focused humanoid robot priced at $20,000. The robot's hands have 22 degrees of actuated freedom, enabling human-like dexterity. Neo operates quietly at 22 decibels and sold out its first-year production of 10,000 units within five days. The company emphasizes home deployment over factory work, believing home environments generate richer AI training data. Self-laundry and dressing are still in active development — full autonomy remains years away.

Zoom has launched a batch of new AI tools at its Zoomtopia conference, including an AI assistant that can take notes, schedule meetings on Zoom and rival platforms like Google Meet and Microsoft Teams, and search across Google and Microsoft services. It also introduced photorealistic avatars, live translation, and custom AI agents. These features aim to make video calls more efficient and automated.

Nvidia has partnered with Microsoft to launch new laptops featuring its RTX Spark graphics cards, designed to bring AI capabilities to everyday computing tasks. The RTX Spark GPUs aim to make AI features like real-time image generation, video editing enhancements, and intelligent assistants more accessible on standard Windows laptops without requiring expensive workstation hardware.

Anthropic, the AI company behind Claude, has officially filed paperwork with U.S. regulators to go public — a major step toward what could be one of the biggest tech IPOs ever. The company was valued at $965 billion after raising $65 billion just last week. Investors like Sequoia, Capital Group, and Singapore's sovereign funds GIC and Temasek backed the latest round, betting big on the AI race against OpenAI. The filing kicks off a process that will let everyday investors buy into the company for the first time.

Anthropic, the AI company behind Claude, has officially filed paperwork with U.S. regulators to go public — a major step toward what could be one of the biggest tech IPOs ever. The company was valued at $965 billion after raising $65 billion just last week. Investors like Sequoia, Capital Group, and Singapore's sovereign funds GIC and Temasek backed the latest round, betting big on the AI race against OpenAI. The filing kicks off a process that will let everyday investors buy into the company for the first time.

Singapore-based robotics firm Sharpa is providing advanced robotic hands for a humanoid robot project with Nvidia and Chinese robot maker Unitree. The H2 Plus robot will be available to researchers in late 2026, enabling them to train AI for precise tasks like assembly or administering IV drips. This makes sophisticated humanoid research more accessible to universities worldwide.

Anthropic has confidentially filed for an IPO with the SEC, beating OpenAI to the public market. The Claude AI maker's move highlights the fierce competition for investor dollars in AI.

1X, a robotics company, is developing Neo — a home-focused humanoid robot priced at $20,000. The robot's hands have 22 degrees of actuated freedom, enabling human-like dexterity. Neo operates quietly at 22 decibels and sold out its first-year production of 10,000 units within five days. The company emphasizes home deployment over factory work, believing home environments generate richer AI training data. Self-laundry and dressing are still in active development — full autonomy remains years away.

Nvidia has partnered with Microsoft to launch new laptops featuring its RTX Spark graphics cards, designed to bring AI capabilities to everyday computing tasks. The RTX Spark GPUs aim to make AI features like real-time image generation, video editing enhancements, and intelligent assistants more accessible on standard Windows laptops without requiring expensive workstation hardware.

Senator Elizabeth Warren is urging President Trump to close a regulatory loophole that currently allows American AI chips to be shipped to foreign subsidiaries and affiliates of Chinese companies. The concern is that these chips—which include advanced Nvidia products—could end up being used by Chinese entities despite existing export restrictions targeting state-owned enterprises directly. Warren argues the current rules allow Chinese firms to access restricted US technology through their overseas operations.

Singapore's GovTech agency is building a registry to track AI agents used by 150,000 public officers, ensuring these automated tools operate securely without leaking sensitive government data. The "AI Assistant Desk" suite includes customizable security rules and automated hygiene checks to prevent offensive language or unauthorized actions. Officials are already using AI to code, draft reports, and process business licenses — and AI-assisted pen testing is now scanning 2,000 government systems for vulnerabilities.

The European Union is drafting new cloud rules that would restrict Big Tech companies like Amazon, Microsoft, and Google from bidding on sensitive government cloud contracts. The aim is to prevent these giants from dominating critical digital infrastructure. These rules could reshape how European governments spend billions on cloud services and push more business toward smaller, European-owned providers.

Hackers are exploiting Meta's AI support bot on Instagram to gain access to high-profile accounts. They simply ask the chatbot to change the account's email address, bypassing security. The exploit has been active for months, leading to hijacked accounts like Barack Obama's. Meta says it has now patched the flaw, but the damage shows how AI customer service tools can be dangerously naive.

Chinese AI startup MiniMax launched its M3 model and saw shares jump 54% on its Hong Kong Stock Exchange debut. The company is a notable competitor in China's AI race, but has no direct link to Singapore.

Zoom has launched a batch of new AI tools at its Zoomtopia conference, including an AI assistant that can take notes, schedule meetings on Zoom and rival platforms like Google Meet and Microsoft Teams, and search across Google and Microsoft services. It also introduced photorealistic avatars, live translation, and custom AI agents. These features aim to make video calls more efficient and automated.