Agents drive adoption; EU AI Omnibus in force

Anthropic / Claude ecosystem

Egencia AI connector in Claude launches for business travel

American Express Global Business Travel's (Amex GBT) Egencia AI connector has launched in Claude, marking the first end-to-end agentic integration in managed travel. This allows for full transaction completion (search, book, manage) within an AI assistant using the Model Context Protocol (MCP) for agent-to-agent reasoning.

Frontier model providers

No significant new developments.

AI developer tooling & infrastructure

Orchestrate Agentic Workflows with North Automations

Cohere has launched North Automations, an enterprise workflow orchestration platform designed to coordinate AI agents at scale with built-in governance. This platform targets a $550 billion market opportunity in managing complex AI-driven processes.

https://devops.com/github-gives-teams-more-control-over-copilots-cloud-agent-in-linear/

GitHub's Copilot cloud agent integration with Linear has reached general availability. New controls allow for model selection, custom agents, branch targeting, and mid-session steering directly from the Linear planning tool.

https://www.arcade.dev/blog/why-were-still-betting-on-mcp/

Arcade.dev has announced its continued strategic investment in the Model Context Protocol (MCP) following the July 28 specification release, which makes MCP stateless and enterprise-ready. This commitment reinforces MCP's role in facilitating robust AI agent interactions.

You.com | The You.com MCP Server Comes to Warp

The You.com MCP Server now integrates with Warp terminal agents, providing live-web access without model-gating through four tools: search, contents, research, and finance. This integration operates over the standard Model Context Protocol (MCP).

Cloud & platform providers

AWS Weekly Roundup: Local Zone in Athens, Claude Opus 5 on AWS, Lambda durable execution for .NET, and more (July 27, 2026)

AWS now offers Anthropic's Claude Opus 5 via Bedrock with zero data retention. Additionally, Lambda durable functions for .NET are generally available, and a new Local Zone in Athens enables latency-sensitive workloads in Greece.

Cloudflare open-sources a debugger for privacy protocols used by Apple and Microsoft, with AI agents in mind

Cloudflare has released an open-source debugger for privacy protocols, specifically those used by Apple and Microsoft, with AI agent development and verification as a primary consideration. This tool is designed to support the creation and auditing of privacy-preserving AI applications.

Fastly gains Experian agent checks as 53% of web traffic turns automated

Fastly has extended Experian's Agent Trust decisioning layer to the network edge, enabling real-time AI agent verification and transaction authorization before requests reach origin servers. This integration addresses the rise of automated web traffic, now accounting for 53% of all web requests.

AI policy, regulation & governance

AI chatbots citing Russian propaganda sourced from EU-sanctioned outlet

Researchers from UK think tank Demos discovered that five mainstream large language models (xAI Grok, OpenAI GPT, Anthropic Claude, Google Gemini, Mistral) are citing content from an EU-sanctioned Russian propaganda outlet, r-FBI, as legitimate sources. This amplifies disinformation regarding Ukraine.

AI Omnibus enters into force

The European Union's AI Omnibus regulatory package came into force on July 27, 2026. This package simplifies EU AI rules while maintaining safeguards and providing extended compliance timelines for high-risk AI systems.

AI Workplace Safety Guidance: What Builders Should Know

Safe Work Australia has issued new guidance, signaling that AI and digital tools are now considered a work health and safety matter. This foreshadows potential harmonization of rules across Australian states, following NSW's 2026 digital work systems legislation.

Industry & market moves

Cognizant and Anthropic expand partnership to embed Claude in Cognizant's industry platforms, helping clients close the gap between AI promise and business outcomes

Cognizant has become a Global Premier Partner in Anthropic's Claude Partner Network, integrating Claude across its engineering platforms. This expansion aims to deliver production AI results in manufacturing, life sciences, and insurance, with a goal of scaling a 30,000+ Claude-certified workforce.

Ode with Anthropic and LogicGate Announce Partnership to Scale Expert-Led Client Outcomes through Accelerated Agentic Capabilities

Ode, backed by Anthropic, Blackstone, and other institutional investors, is partnering with LogicGate to embed frontier AI models into enterprise Governance, Risk, and Compliance (GRC) workflows. This collaboration aims to accelerate agentic capabilities and scale expert-led client outcomes.

South Australia secures up to 1 GW of AI data centre infrastructure

1414 Degrees and an undisclosed Australian data centre operator have signed an agreement to develop up to 1 GW of AI compute infrastructure at the Aurora Energy Precinct near Port Augusta, South Australia. This initiative is backed by 900 MW of solar generation and battery storage.

Ilya Sutskever’s Safe Superintelligence partners with Nvidia to scale its AI research

Safe Superintelligence (SSI) has emerged from two years of stealth to announce a $5 billion partnership with Nvidia. This investment grants SSI access to Nvidia's Vera Rubin GPUs, enabling a tenfold scaling of its alignment-focused AI research infrastructure.

Enigma raises $70M to make controlling a robot as easy as adjusting the volume

Enigma, a robotics AI research lab, has raised a $71 million seed round to develop intuitive human-robot interfaces. Their approach focuses on large-scale human-interaction experiments, diverging from competitors solely focused on model capabilities.

AI product & feature launches

Oracle AI Agent Studio Opens to Pro-Code Devs

Oracle has extended its AI Agent Studio to professional developers, offering support for VS Code, Claude Code, and Codex. This completes Oracle's three-tier builder stack (no-code, low-code, pro-code) for developing autonomous enterprise software within Fusion Cloud Applications.

Introducing Atlas: Wiz's AI vulnerability researcher

Wiz has launched Atlas, an AI system that ranks first on the CyberGym benchmark. Atlas has uncovered over 200 zero-day vulnerabilities in heavily audited open-source projects through autonomous adversarial validation.

IAG bets on agentic AI where conduct risk is highest

IAG (Insurance Australia Group) is deploying customer-facing agentic AI in its claims handling, a high-risk area for conduct. This move occurs precisely where regulatory bodies ASIC and APRA are increasing their enforcement oversight.

Celeris Unveils Celeris-1: Unlocking Real-Time AI Through Diffusion-Based Language Generation

Celeris has unveiled Celeris-1, a diffusion-based language model that delivers near-GPT-5 reasoning capabilities at 15 times faster speed. This advancement aims to enable real-time AI applications.

Gem Browser imagines an internet where every website is yours to redesign

Google DeepMind's Gem Browser allows users to utilize Gemini AI to redesign any real website's interface on their own screen, without altering the publisher's original page. This empowers individual customization of web experiences.

Microsoft launches its first cybersecurity model, plus a new agentic cybersecurity system

Microsoft has launched its first specialized cybersecurity AI model, MAI-Cyber-1-Flash, alongside an agentic security platform called Perception. This system is designed to automate vulnerability detection and remediation at an enterprise scale.

Siemens and Nvidia launch self-verifying agentic AI for chip and circuit board design

Siemens and Nvidia have partnered to integrate self-verifying agentic AI workflows into semiconductor and PCB design tools. This Fuse EDA AI Agent system, combined with Nvidia AI technology, enables trusted autonomous agents to validate decisions against physics-based EDA engines, leading to faster and more reliable engineering outcomes.

Research with immediate practical relevance

Physics Could Transform Computing and AI forever

New research shows that physics-based AI systems can perform image generation tasks with approximately 1000 times lower energy consumption than conventional neural networks, while matching the performance of earlier-generation deep learning models.