Agentic AI

Meta launches Muse Spark 1.1, an upgraded AI model targeting agentic tasks and coding assistance. This move intensifies competition with OpenAI and Anthropic. The model is now publicly accessible, with competitive pricing and a focus on Meta’s ecosystem. While Meta continues open-source efforts, it’s prioritizing commercial access to its advanced AI, aiming to demonstrate ROI on significant AI investments and catch up with leading AI models.

Salesforce acquired AI customer service platform Fin for approximately $3.6 billion. This move significantly boosts Salesforce’s Agentforce platform with advanced generative AI agent capabilities, allowing autonomous customer service across various channels. The deal signals a major acceleration in enterprise adoption of agentic AI solutions, intensifying competition in the tech sector.

Mistral AI, a young French startup, is rapidly emerging as a major contender in the AI landscape. The company is developing advanced AI models and investing in infrastructure, aiming to position Europe as a key player against US dominance. Mistral’s focus on agentic AI, enterprise deployment, and vertical integration, including potential in-house chip development, signals ambitious growth and a strategic vision for the future of AI.

Microsoft’s new Majorana 2 quantum chip boasts a thousandfold increase in qubit reliability and a 20-second qubit lifetime, dramatically advancing the path to commercially scalable quantum computing by 2029. This breakthrough was significantly enabled by Microsoft Discovery, an agentic AI platform that accelerated R&D by optimizing fabrication, automating measurements, and synthesizing vast research data. The platform is now available to enterprise clients, promising to revolutionize scientific discovery across industries.

The generative AI race is heating up in the coding assistant market. Anthropic leads with Claude Code, while OpenAI, Google, and Microsoft are strategically integrating AI into their cloud ecosystems. Google highlights agentic AI with Antigravity 2.0 and Gemini 3.5 Flash. Microsoft will launch a cost-effective Copilot model, focusing on enterprise. This rapidly growing market, projected to reach $30 billion by 2031, sees competition fostering innovation and driving developer adoption.

Okta beat fiscal Q1 expectations, with revenue up 11% driven by strong demand for identity security amid the rise of agentic AI. The company reported adjusted EPS of 91 cents on $765 million revenue. CEO Todd McKinnon highlighted AI’s role as a growth catalyst, emphasizing Okta’s long-term strategy to provide foundational security infrastructure for future AI deployments. Okta is enhancing offerings like “Okta for AI agents” to address security needs arising from AI expansion.

Mistral AI is exploring custom chip design and manufacturing for cost reduction and infrastructure control, aiming to compete with U.S. AI giants. The company is investing heavily in data center expansion in France and Sweden to boost compute capacity and launching an enterprise agent platform called “Vibe” for complex task execution. Mistral AI aims for significant revenue growth by 2026.

Embodied AI, moving beyond digital to physical realms like warehouses and public spaces, introduces new safety risks. Singapore’s IMDA released a Model AI Governance Framework for Agentic AI, stressing stringent controls, monitoring, and human oversight for autonomous systems interacting with the physical world. This shift necessitates regulatory paradigms akin to aviation and critical infrastructure, focusing on reliability, operational safety, and iterative testing. Accountability remains with human operators, guiding complex AI ecosystems from development to deployment.

OpenAI establishes its first Applied AI Lab outside the US in Singapore, investing S$300 million and creating over 200 technical roles. Singapore also unveils an updated agentic AI governance framework, building on previous iterations and incorporating feedback from over 60 organizations. This framework offers clearer guidance on the responsible deployment of AI agents, addressing risks of multi-agent systems and human accountability, with detailed case studies illustrating practical implementation.

Google unveiled Gemini 3.5 Flash, a faster, more affordable AI model, at its I/O conference. They also introduced Gemini Spark, an AI agent for task automation, and Omni, a world model for simulating physical environments and advanced video editing. These moves emphasize Google’s commitment to AI innovation and integrating AI into user-facing services.