INVA-AI-HERALD - Google Launches a Universal Gemini Agent for Work
INVA-AI-HERALD
📰 AI NEWS
1. 🧠 GPT-6 Turns ChatGPT Into an Interactive AI Workspace
📰 WHAT HAPPENED !!
OpenAI rolled out GPT-6 in ChatGPT with a new Intelligent UI that can turn answers into interactive experiences. Instead of returning only text, ChatGPT can now compose responses using graphics, charts, forms, tappable controls and small tools such as calculators or games. GPT-6 can also begin answering while it continues reasoning or using tools, allowing users to start reading before the complete response is finished. The rollout began globally for Plus, Pro, Business and Enterprise users, while Free and Go access follows through the broader rollout. The change pushes ChatGPT toward becoming an interactive computing environment rather than a text-only assistant.
🧠 WHAT CAN IT DO?
GPT-6 can combine text, visuals and interactive components in one response. It can create diagrams, charts, forms, calculators and simple games, while also beginning a useful response before its full reasoning or tool work has completed.
🌍 REAL-WORLD APPLICATIONS
- Learning: Build interactive explanations and visual lessons.
- Productivity: Create calculators, forms and task tools.
- Research: Combine explanations with diagrams and charts.
- Daily tasks: Turn natural-language requests into usable interfaces.
2. ☁️ Google Launches a Universal Gemini Agent for Work
📰 WHAT HAPPENED !!
Google Cloud unveiled Gemini at Work 2026, introducing a universal Gemini agent for workplace tasks. The agent can answer questions, handle knowledge work, create images and media, and write and run code. Google is also putting Gemini directly inside Gmail, Drive, Docs, Slides, Sheets, Chat and Calendar while preserving shared memory, skills and controls. New analytics capabilities let users ask plain-language questions about business data, while specialized tools and connectors target financial-services and legal teams. Google also outlined identity, authorization, sandboxing and network controls for governing enterprise agents. The initiative is designed to move Gemini from an assistant into a connected workplace agent.
🧠 WHAT CAN IT DO?
Gemini can work across core Workspace applications, use shared context, answer business-data questions and execute knowledge-work tasks. Its enterprise controls are designed to manage identity, authorization, permissions and secure execution as agents become more autonomous.
🌍 REAL-WORLD APPLICATIONS
- Office work: Handle cross-application tasks.
- Analytics: Ask business questions in natural language.
- Finance: Use specialized AI tools and connectors.
- Legal teams: Work with domain-specific knowledge.
3. 💰 Manus Raises $500M+ After Breaking Away From Meta
📰 WHAT HAPPENED !!
Butterfly Effect, the parent company of Manus, completed a funding round of more than $500 million as the AI-agent startup returned to independent operations after its planned Meta acquisition was unwound. The round was co-led by Boyu Capital and IDG Capital, with existing investors including Tencent, HSG and ZhenFund participating. Manus builds agents that can perform real-world tasks rather than simply answer questions. The new financing gives the company fresh capital to continue hiring and product development while it operates independently. The event is notable because it combines a major AI-agent financing with a rare reversal of a previously announced technology acquisition.
🧠 WHAT CAN IT DO?
Manus is built around autonomous agents capable of carrying out multi-step digital tasks for users. Instead of only generating answers, the system can perform actions and coordinate workflows, making it part of the rapidly expanding personal-agent category.
🌍 REAL-WORLD APPLICATIONS
- Research: Perform multi-step information tasks.
- Automation: Execute digital workflows.
- Personal assistants: Handle real-world requests.
- Business operations: Coordinate repetitive online work.
4. 🔗 Upscale AI Launches Token Fabric to Mix AI Chips
📰 WHAT HAPPENED !!
Upscale AI launched Token Fabric, a platform designed to connect AI chips from different suppliers inside the same data center. The system combines networking hardware, software and monitoring capabilities so cloud operators can build mixed-vendor AI clusters without creating separate networking architectures for each accelerator family. Upscale says the platform can identify bottlenecks and flag equipment that may need replacement, helping expensive AI processors remain productive instead of waiting for data. The company is targeting large cloud providers and specialized AI infrastructure operators. Token Fabric is planned to roll out in stages, beginning with chip-to-chip connectivity and expanding toward a broader platform.
🧠 WHAT CAN IT DO?
Token Fabric links processors from multiple AI-chip suppliers and monitors infrastructure for bottlenecks. Its goal is to make heterogeneous AI data centers easier to operate while keeping expensive accelerators supplied with data and compute resources.
🌍 REAL-WORLD APPLICATIONS
- AI data centers: Combine different accelerator families.
- Cloud computing: Improve mixed-hardware deployments.
- Infrastructure monitoring: Detect bottlenecks.
- GPU efficiency: Reduce processor idle time.
5. 🤖 Mecka AI Raises $60M to Teach Robots Human Skills
📰 WHAT HAPPENED !!
Mecka AI raised $60 million in Series B funding led by Sequoia Capital to build the data infrastructure needed to train humanoid and other physical robots. The company collects human-motion data by having people record everyday activities while wearing body sensors and using smartphones. Those recordings are then analyzed to create training data for robots learning how humans move and manipulate objects. Investors include NVIDIA, Microsoft’s M12, Qualcomm Ventures and Samsung, alongside existing backers. Mecka is positioning itself as a robotics-data company, addressing a different bottleneck from the hardware race: giving machines enough real-world human behavior to learn useful physical skills.
🧠 WHAT CAN IT DO?
Mecka captures human motion and behavior as training data for robots. The platform is designed to provide physical-AI systems with richer examples of manipulation and everyday actions, helping developers train machines for real-world environments.
🌍 REAL-WORLD APPLICATIONS
- Humanoid robots: Train physical manipulation skills.
- Industrial robots: Learn real-world human workflows.
- Robot datasets: Build large motion collections.
- Physical AI: Improve machine learning from demonstrations.
6. 🏥 Healthleap Raises $38M to Spot Hidden Risks in Hospital Patients
📰 WHAT HAPPENED !!
Healthleap raised $38 million across seed and Series A funding for an AI platform that flags hospital patients who may have undiagnosed conditions. The company connects to electronic health records and combines structured information such as lab results, vital signs and weights with signals extracted from clinicians’ notes. Its system generates risk scores for care teams to review, rather than making diagnoses itself. Healthleap says its technology is deployed across more than 50 hospitals and is expanding beyond malnutrition and delirium toward additional clinical risks. The financing will support engineering, product development, sales and clinical validation as the platform expands.
🧠 WHAT CAN IT DO?
Healthleap analyzes structured hospital data and clinical notes to surface patients who may need closer review. Its system produces risk signals for care teams, helping prioritize attention without presenting the AI output as a medical diagnosis.
🌍 REAL-WORLD APPLICATIONS
- Hospitals: Identify patients needing closer review.
- Clinical screening: Detect hidden risk signals.
- EHR analysis: Extract useful information from records.
- Care teams: Prioritize limited clinical attention.
7. 🛡️ Rein Security Raises $25M to Guard AI Agents at Runtime
📰 WHAT HAPPENED !!
Rein Security raised $25 million in Series A funding to protect AI agents while they operate inside production environments. Its platform provides real-time visibility into agent behavior, applies guardrails to limit risky actions and helps organizations govern how agents interact with systems. Rein also focuses on supply-chain security for AI agents, an emerging problem as companies connect autonomous software to more tools and credentials. The new funding, co-led by Glilot Capital and Sienna Venture Capital, brings the company’s total raised to $35 million. Rein plans to invest in product development, agentic-security research and global hiring as enterprise agent deployments accelerate.
🧠 WHAT CAN IT DO?
Rein monitors agents as they execute actions, providing behavioral visibility, runtime guardrails and governance. Its platform is designed to reduce risky autonomous behavior while helping security teams understand what agents are doing across production systems.
🌍 REAL-WORLD APPLICATIONS
- Agent security: Monitor autonomous actions.
- Enterprise AI: Add runtime controls.
- Supply chains: Protect agent dependencies.
- Governance: Track agent behavior.
8. 📑 Polygrade Launches AI Claims Platform With $6.3M
📰 WHAT HAPPENED !!
Polygrade launched publicly with $6.3 million in funding to automate warranty claims using AI. The Toronto-based company is building software for warranty providers that goes beyond recording claims or routing tickets: it is designed to help carry out the work needed to process and resolve them. Construct Capital led the round, with American Family Ventures and Nine Four Ventures participating. Polygrade will expand its engineering and commercial teams and plans to move beyond home warranties into extended warranties, OEM warranties and equipment-breakdown insurance. The company is targeting a large administrative workflow where fragmented systems and manual processing can slow claim resolution.
🧠 WHAT CAN IT DO?
Polygrade uses AI to help warranty providers process and resolve claims rather than simply log or route them. Its platform is designed to automate more of the work performed between claim submission and final resolution.
🌍 REAL-WORLD APPLICATIONS
- Home warranties: Automate claim processing.
- OEM warranties: Handle complex warranty workflows.
- Insurance: Reduce manual administrative work.
- Claims teams: Speed up resolution.
🔥 CURRENT AI NEWS
9. 🧬 Rivercell Raises $25M to Build an AI Virtual Cell
.png)
Paris startup Rivercell has emerged with about $25 million to build an AI “virtual cell” capable of modeling how human cells respond to drugs and genetic changes. The company is aiming to simulate biological behavior in software, potentially allowing researchers to explore experiments before running them in the laboratory. Rivercell is developing an indication-agnostic model rather than focusing on a single disease, reflecting the growing effort to create broad biological simulation systems. The approach could eventually help drug developers compare hypotheses and prioritize experiments, reducing the number of physical tests needed during early-stage discovery.
10. 🎙️ Google Releases an Offline AI Meeting Assistant for Mac
Google released AI Edge Foresight, a Mac application designed to capture meeting notes using on-device AI. It can transcribe meetings, combine manual notes with AI-generated material and use local models to answer questions from a knowledge base built from documents. The application uses EmbeddingGemma 2 for multimodal embeddings and Gemma models for local assistance, allowing key functions to work without sending meeting information to the cloud. It gives Google a new position in the competitive AI meeting-assistant market while emphasizing local processing, privacy and real-time access to information.
11. 🌪️ Gudea Raises $7M to Predict Which Internet Stories Will Explode
Gudea raised $7 million in seed funding for an AI platform designed to predict which online narratives are likely to spread widely. Instead of simply counting mentions or hashtags, Gudea maps how information moves across hundreds of platforms and analyzes behavioral patterns. Its system combines deep-learning techniques, including graph neural networks, with language models that summarize findings. The company says its customers use the platform for reputational risk, emerging trends and other situations where understanding a narrative early can matter. The new funding will help expand the company's coverage, team and product capabilities.
12. 🎯 Verso Raises $6M to Turn Consumer Research Into an AI Workflow
Paris-based Verso raised $6 million in seed funding for an AI-powered consumer-intelligence platform that conducts interviews and analyzes responses. The system can manage research from study design and participant recruitment through interviews and automated analysis, giving companies a faster route from research question to usable consumer insight. Verso says its platform can accelerate qualitative research dramatically while reaching more respondents than traditional agency workflows. The company plans to expand internationally, add more consumer-data sources such as quantitative research and social listening, and develop capabilities that allow AI agents to use its research directly.
🧩 AI TOOLS
🐝 OpenSwarm
An AI-first desktop workspace where multiple agents can work in parallel across browsers, apps, and connected tools, letting you orchestrate complex tasks from one place.
🎥 Griffin by Tavus
A real-time human-interaction model that sees, hears, speaks, and responds simultaneously, bringing face-to-face conversational AI closer to natural human interaction.
💧 Liquid Inference
A new AI inference marketplace that dynamically prices model inference, creating an exchange-style way for users to access and compare AI compute.
🛡️ AgentGuard
A lightweight security scanner for AI-agent Skills that checks SKILL.md files for risky capabilities and produces a trust report before you install them.
📞 ClawCall
An AI phone caller that can dial US numbers, navigate phone menus, wait on hold, speak with people, and return the result and available transcript evidence to your agent.
🧩 Semwright
An open-source runtime that gives AI agents structured access to real software through typed operations, permissions, application drivers, and inspectable execution results.
🤖 pmtui
A terminal tool for keeping long-running AI coding sessions moving, using an external loop to refresh prompts and handle routine agent decisions without constant supervision.
🖥️ offstage
An open-source environment that gives computer-use agents their own macOS account for background app testing, keeping the agent’s windows, input, and automation away from your active desktop.
🧑💼 Leanback
An AI assistant for engineering managers that checks team members in through Slack, reads tickets and calendars, detects blockers, and produces a prioritized action plan.
📱 BotBus
A control hub for AI coding agents that lets you monitor, approve, follow up on, and start agent tasks from your phone or watch across connected computers.
🧠 PROMPT OF THE DAY
POV: A Message From 2050 Just Arrived
Important Note: You’ll need to use the prompt below to get the complete result.
What This Prompt Does
“POV” and short cinematic “what if” concepts are a strong fit for AI video because they create an immediate story without needing a complicated setup. This prompt turns one simple idea into a 10–15 second cinematic vertical video with a visual hook, escalating reveal, and satisfying ending.
📋 The Prompt
Create a cinematic vertical AI video for TikTok, Instagram Reels, or YouTube Shorts.
CORE IDEA:
[A strange or impossible event that happens in an ordinary place]
SETTING:
[Bedroom / classroom / street / office / café / other]
MAIN CHARACTER:
[Describe the person or leave as “first-person POV”]
STYLE:
Photorealistic, cinematic, believable lighting, detailed environments, natural motion, realistic physics, immersive camera movement.
VIDEO STRUCTURE:
0–2 seconds:
Start immediately with a visually surprising event that makes the viewer wonder, “What is happening?”
2–8 seconds:
Gradually reveal what is happening through visual action. Keep the camera moving naturally and make the transformation or event easy to understand without narration.
8–12 seconds:
Introduce one unexpected detail that changes the meaning of the scene.
12–15 seconds:
End with a strong visual payoff or mysterious final moment that makes the viewer want to replay the video.
CAMERA:
Use a realistic handheld first-person POV. Begin close to the main action, then reveal the environment with a smooth camera movement. Use natural depth of field and realistic focus changes.
AUDIO:
Create subtle environmental sound effects that match the scene. Build tension naturally. No distracting music unless it improves the story.
IMPORTANT:
- Keep the story understandable without dialogue.
- Use one clear visual idea rather than many unrelated effects.
- Make the movement physically believable.
- Keep character appearance, clothing, objects, lighting, and environment consistent throughout.
- Do not add text, subtitles, watermarks, logos, or random objects.
- Do not imitate a specific creator, movie, or existing viral video.
- Make the concept original and visually surprising.
- Output as a 9:16 vertical short video.🧪 Try It With
Core idea: “A phone on my desk suddenly receives a video message recorded 24 years in the future.”
Setting: Bedroom
Main character: First-person POV
🎯 What You’ll Get
A short cinematic mystery video that hooks viewers immediately, builds curiosity through visual storytelling, and ends with a reveal designed to make the clip feel worth replaying or sharing.
Comments
Post a Comment