Callista AI Weekly

Callista AI Weekly (June 16 - June 22, 2025)

June 23, 202511 min read
Custom HTML/CSS/JAVASCRIPT

This week, the global AI landscape delivered a wave of transformative developments, from hands-on enterprise deployments and new agentic AI tools to regulatory milestones and breakthrough research. For Swiss business leaders, these stories offer a window into how AI is rapidly reshaping industries, compliance requirements, and the very fabric of digital innovation. Here’s your comprehensive briefing on the most impactful AI news and what it means for your organization.

New AI Use Cases

AI is no longer just a buzzword — it’s powering real-world solutions across sectors, from manufacturing and public services to retail and software development. This week’s highlights showcase how businesses are leveraging AI to drive efficiency, unlock new revenue streams, and address operational challenges.

AI-First Strategies for SMEs

  • Freshworks detailed its “AI-first” strategy, focusing on making agentic AI software accessible for mid-sized and small businesses. Its AI Agent Studio and Freddy AI Copilot have already attracted 2,700 paying users, helping drive a 19% revenue increase in Q1 2025. This demonstrates how even smaller organizations can harness AI to boost productivity and customer engagement.

AI in Retail and Manufacturing

  • Yum China (KFC) rolled out “Q-Smart,” an AI-powered hands-free assistant tool in select KFC stores. The system supports restaurant managers with staff scheduling, inventory management, and food quality inspections, aiming to cut wait times and improve delivery speed. This follows similar AI initiatives at Yum Brands in the U.S., highlighting the global trend toward AI-driven operational efficiency in retail.

  • Nvidia & Foxconn are set to deploy humanoid robots at a new Foxconn factory in Houston, which will manufacture Nvidia’s AI servers. This marks the first time Nvidia products will be made with human-like robots, a transformative step for electronics manufacturing. The robots, operational by Q1 2026, are expected to address labor challenges and set new standards for efficiency.

AI-Driven Mobility

  • Tesla began a limited trial of its robotaxi service in Austin, Texas, with autonomous taxis monitored by safety drivers. If successful, this pilot could pave the way for broader rollouts, offering new revenue streams and reduced operational costs for the company.

AI for Developers and Content Creators

  • Oracle & xAI: Oracle added Elon Musk’s Grok 3 models to its cloud platform, enabling enterprise customers to integrate advanced AI for content creation, research, and business process automation—all under Oracle’s enterprise-grade security.

  • Google started training its Veo 3 AI video generator using YouTube content, enhancing video creation capabilities for businesses and creators. This opens new monetization opportunities but also raises questions about the impact on traditional content creators.

  • Y Combinator/Asterisk launched Claudia, a free open-source GUI for developers using Anthropic’s Claude Code. By replacing terminal-based interfaces with visual dashboards, Claudia reduces development complexity and boosts productivity for AI-powered coding projects.

Major Vendor Updates

The world’s leading AI vendors continued to push the boundaries, unveiling new models, features, and strategic partnerships that are reshaping the competitive landscape.

OpenAI

  • ChatGPT Enterprise: OpenAI is shifting from fixed subscriptions to a flexible, credit-based pricing system, with discounts for bundled or multi-year deals. This move makes enterprise AI more accessible and scalable, reflecting a broader industry trend toward consumption-based pricing.

  • ChatGPT Record: OpenAI launched a feature enabling audio capture, summarization, and transcription of meetings and brainstorming sessions, streamlining business communications and documentation.

  • ChatGPT Gov: OpenAI for Government was launched, bringing advanced models to U.S. public agencies. A $200 million pilot contract with the Department of Defense highlights growing government trust in generative AI and sets a precedent for AI adoption in regulated industries.

Google

  • Gemini 2.5 Pro and Flash: Google moved these models from preview to general availability, offering production-ready AI optimized for various use cases and cost structures.

  • Gemini Video Analysis: Google added video upload analysis to Gemini AI on iOS and Android, enabling frame-by-frame video analysis from user prompts without subscriptions—democratizing advanced video tools for businesses and creators.

Anthropic

  • Introduced “remote MCP server” support in Claude Code, allowing the AI assistant to autonomously connect to third-party tools and data for context-aware coding help. This integration boosts developer productivity and security, as Claude can now act on real project data and perform actions, not just chat.

Microsoft

  • Rolled out an AI assistant in Windows 11’s Settings app, allowing users to adjust PC settings via natural language. This agentic AI feature, initially for ARM-based devices, aims to simplify IT support and reduce helpdesk loads for enterprises.

MiniMax

  • The Shanghai-based AI vendor is exploring an IPO in Hong Kong after launching its efficient M1 reasoning model and enhanced text-to-video capabilities. An IPO would mark a significant milestone in the Chinese AI market, offering a new investment vehicle and intensifying competition with Western AI giants.

Other Notable Updates

  • Baidu is on track to open-source its next-gen Ernie AI model by June 30, signalling a shift among Chinese vendors toward open models.

  • Google continues to roll out features from its May I/O announcements, including multimodal Gemini previews and AI summaries in Workspace.

AI Governance Developments

As AI adoption accelerates, governments and regulators worldwide are racing to set guardrails, ensure ethical use, and address societal impacts. This week brought several key developments:

Switzerland

  • The Swiss Federal Office of Energy (SFOE) launched a new survey to assess the electricity consumption and energy efficiency of data centers. While not a new law, this initiative signals future regulatory interest in the energy footprint of AI and data center operations, potentially leading to new efficiency standards or reporting requirements. For Swiss businesses, this is a clear sign to start monitoring and optimizing energy use in digital infrastructure.

European Union

  • The EU AI Act governance rules for general-purpose AI models become applicable on August 2, 2025. The European Commission is considering potential delays due to implementation challenges, but companies developing or deploying high-risk AI systems must prepare compliance frameworks and documentation now.

United States

  • House Republicans advanced the “Big Beautiful Bill,” proposing to override state-level AI regulations for 10 years. While supported by major tech companies, the bill faces bipartisan criticism and could create regulatory uncertainty for businesses operating across multiple states.

  • The U.S. Food & Drug Administration (FDA) rolled out an agency-wide AI system called “INTACT” to modernize operations, streamline regulatory workflows, and improve risk assessments. This marks one of the first enterprise AI implementations in U.S. government, signaling to regulated industries that agencies are embracing AI.

Latin America

  • A dozen countries announced plans to launch “Latam-GPT” in September, the first large-scale AI language model trained for the region’s diverse languages. This open-source model aims to democratize AI access and spur home-grown solutions in education, healthcare, and customer service.

United Arab Emirates

  • The UAE appointed the world’s first cabinet-level “Minister of AI” to lead a National AI System starting January 2026, signalling increased government focus on AI governance and potential new regulatory frameworks.

Global Ethics

  • At a Vatican AI ethics event, Pope Leo XIV urged politicians to ensure AI “functions as a tool for the good of human beings, not to replace them,” especially warning about harms to youth. This high-profile appeal adds moral pressure for ethical AI practices and may influence future policy and corporate governance standards.

Business Communication and Trust

  • A global survey by Sinch revealed that while 97% of businesses plan to use AI in customer communications in 2025, consumer trust varies significantly by generation. Younger consumers are more accepting, but businesses must tailor their AI strategies and be transparent about data usage to build and maintain trust.

Swiss Spotlight: AI Governance and Innovation

Switzerland continues to carve out a unique position in the global AI ecosystem, balancing innovation with a strong focus on ethics, transparency, and compliance.

  • Swiss AI Initiative (EPFL/ETH Zurich): Switzerland is advancing its own large-scale, multilingual, and transparent large language model, trained on over 1,000 languages and developed using the “Alps” supercomputer. This open-source model is designed to reflect Swiss society’s linguistic diversity and data privacy standards, offering local businesses and public services an alternative to foreign AI models.

  • Art Basel and Microsoft launch a Copilot-powered fair app. Visitors aim phones at any artwork; the app returns artist data, price range and a direct chat to the gallery. It is a textbook agentic-AI demo—image recognition triggers a language agent that fetches context-aware answers.

  • CERN Council approves 2026-30 AI strategy. The lab earmarks fresh funds for AI-driven detector upgrades and names a new leadership team with explicit machine-learning mandates. For Swiss suppliers, procurement rounds tied to the High-Luminosity LHC will follow quickly.

  • Swiss Finance Sector: A FINMA survey revealed that 50% of Swiss financial institutions actively use AI, with an average of five applications per institution, and 25% plan implementation within three years. The focus is on compliance, risk management, and process optimization.

Breakthrough Research

AI research continues to push the boundaries of what’s possible, with several breakthroughs this week promising to reshape industries and business practices.

Ultra-Fast Optical AI

  • Researchers at Tampere University & Université Marie et Louis Pasteur demonstrated that AI-like computations can be performed using intense laser pulses in optical glass fibers, achieving near state-of-the-art image recognition thousands of times faster than electronics. This breakthrough in optical computing could lead to a new generation of ultra-fast, energy-efficient AI hardware for high-speed data processing.

Green Cement with AI

  • The Paul Scherrer Institute (PSI) in Switzerland developed an AI model that simulates and optimizes cement formulations in seconds, identifying recipes with significantly lower CO2 emissions while maintaining high mechanical performance. This “digital cookbook” for green cement could rapidly accelerate the construction industry’s decarbonization efforts.

AI Alignment and Safety

  • OpenAI researchers announced a technique to detect and correct “misalignment” in large language models. By monitoring neural activations and retraining on correct data, they could suppress unsafe or unethical outputs, offering an early-warning system for model behavior drift. This is crucial for businesses seeking to ensure their AI systems remain aligned with human values and policies.

AI in Biotech

  • The Weizmann Institute used novel computational methods to design synthetic enzymes far more efficient than those created by previous AI approaches. By combining physics-based modeling with algorithmic design, they engineered enzymes for reactions not found in nature, with the best enzyme about 100× more efficient than earlier AI-designed versions. This could revolutionize manufacturing in pharmaceuticals and green chemistry.

AI Reasoning Limitations

  • A new academic study revealed that even top language models like GPT-4o and Claude 3.5 often guess answers by analogy to past examples rather than reasoning step-by-step, leading to errors on novel problems. This highlights the need for businesses to implement checks when using AI for tasks requiring rigorous logic or novel reasoning.

Parameter-Efficient AI

  • 117 research papers published across five domains introduced new parameter-efficient fine-tuning methods, achieving comparable results with 90% fewer computational resources. This could significantly reduce AI deployment costs for businesses.

AI for Scientific Discovery

  • Microsoft Research announced a breakthrough in computational chemistry using deep learning to dramatically improve density functional theory accuracy, potentially revolutionizing drug discovery and materials science.

AI Coding Agents

  • DeepMind released AlphaEvolve, a coding agent using evolutionary algorithms and LLMs to discover and refine algorithms, already outperforming hand-crafted solutions in multiple tests. This demonstrates AI’s growing capability in foundational computer science research.

Conclusion

The developments of this week paint a vivid picture: AI is not just evolving—it’s actively transforming industries, business models, and regulatory landscapes. For Swiss businesses, the message is clear:

  • AI is everywhere: From retail and manufacturing to finance and public services, organizations are deploying AI to drive efficiency, innovation, and new customer experiences.

  • Agentic AI is maturing: Autonomous AI agents are moving from concept to reality, handling complex tasks and integrating deeply into business processes.

  • Vendor competition is fierce: Major players are racing to offer more powerful, cost-effective, and specialized AI tools, giving businesses a wealth of options but also requiring careful strategic choices.

  • Governance and compliance are critical: Regulatory scrutiny is intensifying, especially around energy use, data privacy, and ethical AI. Proactive compliance and transparent practices will be essential for building trust and avoiding legal pitfalls.

  • Swiss innovation stands out: Switzerland’s focus on multilingual, transparent, and ethical AI—combined with robust infrastructure investments—positions it as a leader in responsible AI adoption.

For business leaders, now is the time to pilot AI projects, invest in upskilling, and build agile strategies that can adapt to the rapid pace of change. The organizations that thrive will be those that embrace AI’s transformative power thoughtfully and proactively.

Ready to explore how Agentic AI can transform your organization? Visit us at https://www.callista.ch/agentic-ai to discover how we can guide your journey into this exciting new era of AI-powered productivity.

Sources

  • Artificial intelligence built for India, in India: NxtGen unveils in-house AI solutions at GPU-powered data centre. The Week

  • Krutrim acquires BharatSahaiyak to scale AI-driven public services across India. ET Government

  • Decoding Freshworks’ AI-first strategy: AI boost for micro-businesses. YourStory

  • OpenAI schraubt am Preis für ChatGPT Enterprise. Netzwoche

  • MiniMax eyeing Hong Kong IPO. Verdict

  • xAI’s Grok models now available on Oracle Cloud. Gulf Business

  • BFE startet neue Erhebung zu Stromverbrauch und Effizienz von Rechenzentren. Netzwoche

  • 97% of businesses to use AI in customer communications in 2025, but are consumers ready for it? PR Newswire

  • AI-like computations performed with intense laser pulses in optical glass fibers. ScienceDaily

  • AI model simulates and optimizes cement formulations for lower CO2 emissions. ScienceDaily

  • Digital media fair: ArtMeta, robots, NFT, and AI at Art Basel. Designboom

  • Reuters, Windows Central, Anthropic, FDA Press Release, OpenAI blog, and other sources as referenced in the research reports.

Back to Blog

General Guisan Strasse 8

6300 Zug, Switzerland

Balkanska 2

Belgrade, Serbia

Contact

+41 43 540 56 85

Newsletter Sign Up

© 2025 Callista

© 2025 Callista