Google Cloud Next 25: Flurry of New AI Capabilities Announced

Also, Anthropic rolls out $200 / mo subscription tier

Author

Jack Lajoie
April 10, 2025 • read time ~ 9 minutes

⚡️ Headlines

🤖 AI

OpenAI launches 'Pioneers Program' to support startups in tuning domain-specific AI models - The program offers select companies collaboration on building custom benchmarks and fine-tuned models to solve real-world problems. [OpenAI]

Drafted uses AI and video resumes to help early-career professionals land jobs - Startup Drafted is leveraging AI to analyze candidates' video resumes and match them with job opportunities more effectively. [TechCrunch]

8Flow wants to map your enterprise workflows with AI and has raised $10M to do it - The startup aims to make enterprise data and processes AI-friendly with a visual interface backed by new funding. [VentureBeat]

Samsung and Google Cloud expand partnership to bring Gemini AI to Ballie home robot - The smart assistant Ballie will now be powered by Gemini AI, enabling more natural interactions and intelligent features. [Samsung Newsroom]

OpenAI’s GPT helps spammers send 80,000 emails that bypassed filters - Security researchers reveal how GPT-powered spam campaigns used sophisticated prompts to evade traditional spam detection. [Ars Technica]

Apple fumbled Siri’s AI makeover as internal conflicts delayed generative tech rollout - Apple reportedly struggled to implement a meaningful AI upgrade to Siri due to internal divisions and missed opportunities, putting it behind rivals like Google and OpenAI in the generative AI race. [The Information]

AI insurtech Ominimo bags first investment at $220M valuation - The AI-powered insurance startup secured funding to expand its automated risk assessment and underwriting platform. [TechCrunch]

WordPress.com launches a free AI-powered website builder - The new tool uses generative AI to help users create websites from scratch using natural language prompts. [TechCrunch]

🦾 Emerging Tech

Crypto markets rebound as memecoins, AI, and DeFi surge amid easing tariff fears - Digital assets recover with renewed interest in AI-linked tokens and decentralized finance. [CoinDesk]

Tariffs and trade tensions could boost Bitcoin adoption, says Grayscale - Grayscale predicts increased Bitcoin demand as global trade uncertainty pushes investors toward decentralized stores of value. [CoinDesk]

🔬 Research

New AI tool mines old MRI scans for insights into multiple sclerosis progression - Researchers introduce MindGlide, a deep learning model that enables analysis of brain lesion and volume data from previously unusable scans. [Nature Communications]

⚖ Legal

Europe outlines vision for AI innovation and competitiveness in new strategy paper - The European Commission's "AI Continent" plan promotes AI development while emphasizing ethical and global leadership goals. [European Commission]

OpenAI attorneys ask court to block Elon Musk from interfering with company operations - The legal filing seeks to enjoin Musk from further “unlawful and unfair” actions that threaten OpenAI’s functioning. [TechCrunch]

Congress reintroduces No Fakes Act to regulate AI deepfakes with bipartisan support - The proposed law targets unauthorized digital replicas of real individuals to combat misuse in music, film, and media. [Billboard]

Congress close to passing deepfake law even as Trump says he wants to use the tech - Lawmakers move toward enacting regulation on AI-generated likenesses while political figures weigh its strategic use. [Ars Technica]

🎱 Random

Black Mirror season 7 review: tech tales with a more human heart - The latest season leans into warmer, more emotional narratives while still probing tech’s darker edges. [The Verge]

🔌 Plug-Into-This

At Google Cloud Next ’25, Sundar Pichai unveiled sweeping integrations of Gemini AI across Google Cloud, Workspace, and cybersecurity offerings—positioning the model as both a development co-pilot and enterprise backbone. The keynote emphasized AI agents, cross-platform orchestration, and a new developer stack grounded in Gemini 1.5 Pro’s extended context window.

  • Gemini 1.5 Pro is now embedded across Workspace (Docs, Gmail, Sheets) and Vertex AI, enabling 1M-token context windows for in-app reasoning and document synthesis.

  • A new “Gemini Code Assist” replaces Duet AI, offering upgraded coding suggestions, debugging, and enterprise-grade security integration.

  • “AI Agents” built on Gemini will perform multi-step workflows and integrate with third-party APIs, previewing a future of autonomous operations across business apps.

  • The Gemini API in Vertex AI now supports grounding via enterprise data and integrates with Google Search to reduce hallucinations.

  • Chronicle security tools now feature Gemini-driven threat detection, summarization, and automated response recommendations.

🧠 Google's shift from discrete AI features to agent-based infrastructure marks a strategic inflection: the company is betting that deeply-integrated, multi-modal models will become core to enterprise operations—not just enhancement layers.

Anthropic is escalating its competitive posture with OpenAI by releasing a Claude Team plan offering access to its top-tier Claude 3 Opus model, targeting businesses and power users with enhanced performance and collaboration features. The plan costs $30 per user/month and emphasizes shared memory, longer context, and safety controls.

  • Claude Team subscribers gain access to 200K-token context windows and early feature previews for Claude 3 Opus.

  • The subscription enables team-wide prompt sharing and memory persistence, allowing collaborative workflows and more consistent interactions.

  • Fine-tuned moderation and usage analytics are included, catering to compliance-conscious organizations.

  • Anthropic positions Claude 3 Opus as "highly capable" in reasoning and multilingual tasks, aimed squarely at enterprise-grade deployments.

  • Mobile apps and custom model personalities are part of the roadmap, widening accessibility and control.

🧪 Anthropic’s deliberate pacing—prioritizing reliability and interpretability—could become a competitive asset as enterprises weigh trust and traceability over raw model horsepower. This could be their inroad to the workplace.

Adobe is integrating generative AI agents into its flagship creative tools, announcing context-aware "Actions" in Photoshop and Premiere Pro that let users instruct AI to perform multi-step edits and tasks in natural language. These upgrades aim to make complex workflows more intuitive and semi-autonomous.

  • Users can now type commands like “make this look like a sunset scene” to trigger chained editing steps inside Photoshop and Premiere.

  • These AI agents understand creative context and can handle sophisticated tasks like background replacement, audio leveling, and scene matching.

  • Adobe Firefly remains the generative engine, but Actions provide a new interaction layer bridging prompts with visual editing logic.

  • Premiere Pro will also get AI-based script alignment, object addition/removal, and real-time shot suggestions.

  • These tools are being rolled out to beta users with wider release later in 2025.

🎨 Adobe’s shift from isolated generative features to task-level agency lines with the broader trend Google is also displaying. Creative software is evolving from a fun toolset to a creative collaborator, in theory, compressing the gap between concept and execution.

 🆕 Updates

📽️ Daily Demo

🗣️ Discourse