January 2026 Summaries

Multi-agent AI systems, where multiple specialized agents collaborate to complete complex tasks, require more than traditional orchestration frameworks to function effectively. Users expect these systems to break down tasks and coordinate seamlessly, providing visible progress and allowing for real-time intervention. However, conventional orchestrators often act as bottlenecks, limiting the fluidity and responsiveness of these systems. To overcome this challenge, a robust AI transport layer is essential, supporting persistent, multiplexed communication and state synchronization across agents. This infrastructure allows for direct, concurrent updates from multiple agents, enhancing transparency and reliability without relying solely on a single orchestrator. The ability to handle mid-task changes and ensure that system activity is visible to users builds trust and ensures that agents can coordinate effectively, ultimately transforming isolated model calls into a cohesive and collaborative user experience.

In discussions with CTOs, AI engineering leads, and product leaders from various industries, it emerges that while the development of agentic AI is progressing, the key challenges lie not in the models themselves but in user experience (UX). As companies transition from Gen-1 to Gen-2 AI, which involves more complex interactions like assistants and agents, the importance of seamless UX becomes apparent. Issues such as latency, communication standards, and continuity across sessions are critical, as users expect progress visibility and the ability to interrupt or steer AI tasks. Despite advancements in models and frameworks, the lack of standardization in agent-to-client communication means many teams must develop bespoke solutions, which complicates scalability. Failures, particularly partial ones, are common, prompting teams to focus on retries and durable workflows. Additionally, cost optimization and presence awareness are becoming priorities to manage AI workloads efficiently. Ultimately, the future success of agentic AI will depend on delivering reliable and user-friendly experiences, rather than merely having the most advanced models.

Amber Dawson explores the complexities of building multi-agent AI systems, using a travel planning demo to illustrate the need for robust infrastructure to support coordination, visibility, and responsiveness. The demo involves an orchestrator and specialist agents for flights, hotels, and activities, highlighting the importance of synchronizing state updates and providing users with real-time visibility into each agent's activities. It emphasizes that while streaming tokens is common, users require deeper insights into agent behaviors, which necessitates infrastructure capable of handling multiple publishers and structured state updates. Ably AI Transport is highlighted as a solution offering reliable transport layers, state synchronization, and presence detection, ensuring that agents remain responsive and synchronized regardless of network fluctuations or complex user interactions. The patterns used in the demo apply to broader enterprise scenarios, including research, document generation, and customer support, demonstrating the critical need for infrastructure that supports concurrent agents, shared state, and user visibility.

Realtime steering in AI systems allows users to dynamically interact with AI assistants, enabling actions like interrupting, redirecting, and guiding responses without losing context or restarting sessions. This capability is increasingly expected by users who desire more natural, human-like interactions with AI, moving beyond static, one-turn exchanges. However, implementing realtime steering is challenging due to the limitations of traditional communication models like stateless HTTP, which struggle with mid-stream redirection and maintaining shared conversational state. Essential components for supporting realtime steering include a transport layer that facilitates low-latency, bidirectional messaging and session continuity, ensuring that control signals reach the AI system efficiently and maintain conversational context across devices. Ably AI Transport provides a solution by offering a transport layer that supports these requirements, allowing developers to build responsive, controllable AI interactions without overhauling their existing infrastructure.

Anticipatory customer experience (CX) represents a transformative shift in how businesses engage with customers, leveraging real-time data infrastructure to predict and address needs before friction occurs. This proactive approach moves beyond reactive support by utilizing continuous data streams, event-driven patterns, and predictive signals to foresee and mitigate issues. Companies like Doxy.me have effectively implemented anticipatory CX by using real-time platforms to identify potential confusion and intervene before it escalates, resulting in significant cost reductions. This strategy is becoming increasingly critical as consumers expect brands to preemptively meet their needs, and industries such as healthcare, retail, and financial services are leading the way. The core of anticipatory CX lies in reliable, adaptive, event-driven architectures that allow businesses to maintain trust and flexibility while scaling to meet evolving customer expectations. Ably's real-time platform exemplifies this by providing the necessary infrastructure for continuous data flow, low-latency communication, and resilience, enabling companies to deliver foresight at scale.

Ably AI Transport, launched by Faye McClenahan, offers a drop-in realtime delivery and session layer that enhances the continuity of AI experiences across device changes, refreshes, and reconnects without requiring an architecture overhaul. Traditional HTTP streaming methods often falter with stateful AI user experiences, as they fail to maintain continuity amid normal user behaviors like network changes or device switches, leading to issues such as partial outputs and duplicated messages. Ably AI Transport addresses these challenges by providing a durable, bi-directional session layer, allowing agents to publish outputs into a session channel accessible from any device, ensuring ordered delivery, and enabling features like live steering and multi-agent coordination. This model and framework-agnostic solution creates a resilient AI user experience by supporting resumable, ordered token streaming, multi-device continuity, and seamless human takeovers, while also incorporating identity and access control, and observability for diagnosing production issues. Through this innovation, Ably AI Transport allows users to maintain continuous, stateful interactions with AI applications, ensuring that conversations are user-scoped rather than tied to specific connections, thus providing a robust, scalable solution for modern AI-driven environments.

Ably has embraced an "AI-first" culture by integrating artificial intelligence into every aspect of its operations, from engineering and marketing to sales and finance. This transformation was driven by a company-wide mandate to explore AI's potential to enhance effectiveness and uncover new possibilities, resulting in a shift from asking if AI can help to assuming it can. The integration is built on three pillars: internal AI adoption to automate tasks and enhance productivity, improving the AI developer experience to make Ably's platform more accessible, and proactive AI product enhancement to meet real customer needs. A crucial development in this transition is the Model Context Protocol (MCP), which links over 140 tools, allowing seamless access to shared data and context for various tasks, thus eliminating repetitive context-switching. Teams have adopted Claude as the primary AI tool, while workflow automation is managed by Relay, but flexibility in tool choice is encouraged to maintain infrastructure compatibility with any AI provider. The cultural shift has resulted in increased productivity and creativity across teams, with AI being used for everything from coding and marketing analysis to sales lead generation and financial reconciliation, all while fostering a mindset of continuous improvement and problem-solving through AI.

The new Ably dashboard provides developers with comprehensive real-time visibility into their applications, enabling them to monitor and debug systems directly from their browser without the need for external tools or setups. This modernized interface offers streamlined navigation and a product-first structure, with dedicated sections for Pub/Sub, Chat, Spaces, LiveSync, and LiveObjects, enhancing usability and readability even during high-traffic periods. Key features include live inspectors that offer real-time insights into data flow, connection, and channel activities, as well as a log search function for querying historical events. Additionally, a new Web CLI allows developers to test Ably's capabilities instantly, and the reporting section provides aggregated insights into application usage patterns. This release marks a significant step toward achieving complete real-time observability, aiming to help developers understand, debug, and optimize their systems more efficiently.