AI in Marketing

Claude Managed Agents has launched a suite of composable APIs designed to expedite the development and deployment of cloud-hosted agents, reducing the time from prototype to production from months to days. This new offering from the Claude Platform simplifies the process by handling infrastructure complexities such as secure sandboxing, authentication, and tool execution. Developers can focus on defining tasks and outcomes while the platform manages orchestration, context, and error recovery. Managed Agents supports both single-task runners and complex multi-agent pipelines, allowing for autonomous long-running sessions and multi-agent coordination. The platform is currently in public beta, with some features available in research preview. Companies like Notion, Rakuten, Asana, Vibecode, and Sentry have already utilized Managed Agents to accelerate their workflows. Notion integrates Claude directly into workspaces for task delegation, while Rakuten deploys enterprise agents across various departments. Asana's AI Teammates enhance project collaboration, and Vibecode uses Managed Agents for rapid app deployment. Sentry's integration allows developers to address bugs more efficiently. Managed Agents is priced based on consumption, with standard Claude Platform token rates and additional charges for active runtime. Developers can access the service through the Claude Console or CLI, with onboarding available via the Claude API. This innovation aims to streamline agent deployment, enhancing productivity and reducing operational overhead.

Next.

Anthropic, an AI company, faces conflicting court rulings over its designation as a supply-chain risk by the Pentagon. A Washington, DC, appeals court upheld the designation, opposing a San Francisco judge's earlier decision to remove it. The designation, under two supply-chain laws typically used against foreign entities, restricts the Pentagon from using Anthropic's AI, Claude, in military projects. The Washington court emphasized the need to avoid disrupting military operations, despite potential financial harm to Anthropic. The San Francisco judge had accused the Department of Defense of acting in bad faith, influenced by Anthropic's restrictions on AI use and criticism of those limits. Anthropic argues the designation is unlawful and claims it has lost business as a result. The company insists its AI lacks the accuracy for sensitive military operations, such as drone strikes, without human oversight. Experts suggest Anthropic has a strong case, but courts often defer to the executive branch on national security. The final rulings on Anthropic's lawsuits could take months, with the Washington court set to hear arguments on May 19. Details on the Pentagon's use of Claude or its transition to other AI tools remain limited.

Meanwhile.

TechCrunch reports that the AI agent Poke, developed by The Interaction Company of California, simplifies automation through text messaging platforms like iMessage, SMS, and Telegram. Launched in March, Poke acts as a personal assistant, managing tasks such as daily planning, calendar management, health tracking, and smart home control. Unlike more complex systems like OpenClaw, Poke is user-friendly, requiring no software installation. Users can create and share automations in plain text, enhancing its functionality. Poke's development was influenced by beta testers who used an earlier email-focused AI assistant for broader tasks. The startup, backed by Spark Capital and General Catalyst, recently secured an additional $10 million in funding, bringing its valuation to $300 million. Poke utilizes various AI models, distinguishing itself from competitors tied to specific providers. While WhatsApp support is limited due to Meta's restrictions, ongoing antitrust probes in the EU, Italy, and Brazil may change this. Poke's ease of use and adaptability aim to make AI-driven automation accessible to a wider audience, addressing security and technical challenges associated with other systems.

On a different note.

Microsoft has terminated an account linked to VeraCrypt, a widely used encryption software, raising concerns about the future of its Windows updates. This action underscores the complexities of the supply chain in open-source software distribution, particularly when such software depends on large tech companies. VeraCrypt's developer shared this development with 404 Media, emphasizing the potential impact on the software's updates. The situation highlights the challenges faced by open-source projects that, even indirectly, rely on major tech firms for distribution and support. The incident serves as a reminder of the vulnerabilities within the open-source ecosystem, where changes in policy or account status by tech giants can significantly affect software availability and functionality for users worldwide.

With that said.

Artificial intelligence is transitioning from experimental phases to large-scale integration within the travel industry, fundamentally altering how companies operate and compete. Travel companies have historically used AI for isolated tasks like chatbots and automated messaging, but the current focus is on embedding AI into the core infrastructure that supports global travel. This shift aims to create a seamless experience for travelers, where AI tools used for destination research also influence booking and service processes. The challenge lies in integrating AI with legacy systems that were not designed for such technology, requiring adaptations to meet security and compliance standards. Companies are now rethinking their platforms to allow AI to function effectively across interconnected systems. Orchestration of specialized AI agents is crucial, enabling them to handle specific tasks while coordinating with various travel partners. This approach ensures reliable performance and minimizes errors. AI's integration into core systems like reservations and revenue management offers personalization opportunities and operational efficiencies by connecting traditionally separate functions. As AI adoption accelerates, companies that succeed will focus on monitoring, testing, and ensuring AI systems are explainable and reliable. Governance and compliance are essential, especially in a regulated industry, to maintain trust and meet evolving regulations. Amadeus, a key player in this transition, collaborates with tech giants and invests in AI-driven solutions to support this industry-wide shift.

In other news.

Researchers have uncovered that thousands of men are using Telegram groups to buy hacking tools and share nonconsensual images of women and girls. These groups, identified by a European nonprofit, are involved in trading abusive content, including intimate images and child sexual abuse material. Over six weeks, AI Forensics analyzed 2.8 million messages across 16 Italian and Spanish Telegram communities, revealing over 24,000 members posting 82,723 images, videos, and audio files. The content often targets celebrities and influencers, but many victims are ordinary women known to the perpetrators. Telegram, with over 1 billion monthly users, is criticized for insufficient regulation. Despite Telegram's claims of removing millions of pieces of content daily, the platform remains a hub for abusive sharing. AI Forensics documented 13 types of abusive content, with access to channels costing between €20 and €50. Telegram's anonymity and large networks facilitate these activities. Similar issues have been reported globally, with Telegram being a key platform for such abuse. The findings highlight the need for stricter regulation to protect victims.

Elsewhere.

Canva has acquired Simtheory and Ortto, expanding its capabilities in AI and marketing automation. The financial details of these acquisitions were not disclosed. Both companies were founded by Chris and Mike Sharkey, who previously established Stayz, later acquired by Fairfax Media. The Sharkey brothers will assume leadership roles within Canva's AI and marketing technology teams. Simtheory enhances Canva's AI platform by enabling teams to build AI assistants that integrate with various tools and manage tasks. Ortto offers a customer data platform combined with marketing automation, allowing teams to manage communications across multiple channels, including email and SMS. Canva aims to transition from a design tool to a comprehensive system supporting the entire workflow, from idea generation to campaign execution and measurement. These acquisitions are part of Canva's strategy to strengthen its AI and marketing infrastructure. Recently, Canva also acquired Doohly, Cavalry, MangoAI, and MagicBrief, reflecting its ongoing acquisition strategy. Canva ended 2025 with $4 billion in annualized revenue, over 265 million users, and a 20% increase in monthly active users. The company continues to evolve its offerings, aiming to support the entire marketing and content lifecycle through its Canva Grow tool.

Finally.

Artificial intelligence excels in translating and generating multilingual content but struggles with cultural nuances and context, particularly in non-English languages. While AI tools offer speed, scalability, and accessibility, they often miss the cultural depth required for effective communication. Literal translations can result in loss of cultural context, incorrect tone, or unintended meanings, making messages feel unnatural. Organizations relying solely on AI for translation risk inaccuracies, especially in sensitive communications. AI is most effective as a support tool, providing initial translations that human experts refine. For meaningful multilingual communication, organizations should prioritize culturally appropriate translations, collaborate with native speakers, and test AI-generated content. Human insight remains crucial for ensuring messages are respectful and understood.