October 2025 Summaries

At the AI & Digital Health Summit in Basel 2025, hosted by Novartis, pharmaceutical leaders, including representatives from companies like Novo Nordisk, Roche, and Sanofi, discussed the integration of AI in clinical development, highlighting its transformative potential and the challenges in its implementation. Novo Nordisk demonstrated a significant advancement by reducing the time for generating Clinical Study Reports from up to 15 weeks to same-day delivery using AI, with regulators showing a preference for the AI-generated versions due to their completeness and objectivity. The conference emphasized that while AI promises efficiency in drug development, especially in documentation automation and agentic AI, the greatest hurdle remains organizational transformation, as the adoption of AI requires changes in employee roles and workflows, often met with resistance due to concerns over professional identity and career progression. Presenters, including Maxwell Lawson from Novartis and Thomas Brookland from Roche, stressed the need for a co-learning approach between regulators and the industry to foster innovation without compromising compliance, highlighting ongoing debates over regulatory frameworks like the EU AI Act. The summit reinforced the message that successful AI integration in the pharmaceutical sector hinges on balancing productivity gains with stringent compliance, and companies like Ona are developing platforms to help navigate these challenges by providing AI-driven solutions that ensure both efficiency and regulatory adherence.

Ona, formerly known as Gitpod, is a platform designed to enhance the development workflow by integrating robust environments, automations, and AI-driven agents called Ona Agents. These agents are tailored to improve productivity by automating repetitive tasks, providing contextual assistance in code creation and review, and facilitating seamless integration with tools like GitHub, Linear, and Jira. Ona's environments offer isolation and configurability for developers, while the agents use a variety of commands and guidelines to streamline processes such as pull request creation, code review, and documentation writing. The platform encourages developers to adopt a 90/10 approach, where Ona handles most of the workload, allowing humans to focus on refining and finalizing tasks. Additionally, Ona supports parallel task execution, previews for frontend and backend changes, and progressive engagement from simple tasks in a web interface to more complex ones in a desktop IDE, thereby optimizing the development experience and ensuring high-quality output.

Enterprises are increasingly leveraging AI agents to orchestrate large-scale code migrations, addressing the challenge of modernizing extensive and complex technology estates. Historically, rule-based systems like OpenRewrite were used for this purpose, primarily benefiting large tech companies with significant Java-focused codebases by offering deterministic transformations that prioritize predictability and auditability. However, these systems are limited by their reliance on predefined patterns and lack of contextual understanding, making them less effective for custom frameworks and proprietary APIs. AI agents, in contrast, can dynamically analyze context, adapt their strategies, and validate changes through comprehensive testing, providing a more flexible and intelligent approach to code transformation. Platforms like Ona exemplify this shift by enabling agent orchestration within secure, sandboxed development environments, allowing enterprises to execute migrations at scale, such as language upgrades and architectural standardization, while maintaining security and regulatory compliance. This evolution from manual, rule-based methods to automated, agent-driven solutions represents a strategic advancement in transforming code migrations from disruptive projects into routine operations.

AI-driven coding assistants are increasingly being recognized for their potential beyond individual developer support, particularly in handling complex, large-scale enterprise challenges related to maintaining vast, interdependent codebases. These agents can efficiently manage tasks such as dependency upgrades, security patching, and architectural migrations across numerous repositories, tasks that are typically time-consuming for human teams. This capability allows for a new operational model where centralized teams can automate organization-wide concerns, freeing up development teams to focus on innovation. However, deploying these agents in secure enterprise environments necessitates overcoming challenges in secure runtime orchestration, robust identity management, comprehensive integration with enterprise systems, and maintaining governance and compliance. Ona is introduced as a platform specifically designed to facilitate these secure, orchestrated agent migrations, operating within a privacy-first deployment model that integrates seamlessly with enterprise infrastructure. By transforming migration work from disruptive events to routine operations, Ona aims to enable organizations to focus on strategic migration strategies without the burden of managing complex agent orchestration infrastructure.

Cloud development environments (CDEs) have evolved from solving simple compatibility issues to becoming essential for secure and AI-assisted engineering. Coder and Ona, two prominent CDE providers, initially addressed similar challenges such as reproducible environments and scalable infrastructure but have since diverged in their approaches. Coder remains focused on self-managed systems that offer flexibility and require user control, making it ideal for teams with strong platform engineering capabilities. Conversely, Ona has shifted to a vendor-managed model that simplifies operations and focuses on secure automation, allowing teams to concentrate on engineering tasks without managing the underlying infrastructure. This difference reflects a philosophical split: Coder offers control and customization, whereas Ona provides ease of use and minimal operational overhead. The platforms also vary in their approach to AI-assisted development, with Coder requiring user-defined configurations and Ona offering built-in, secure agents. Ultimately, organizations must choose between managing complexity themselves or opting for a service that provides rapid evolution and integration, with Coder suiting teams seeking full-stack control and Ona appealing to those desiring a streamlined, secure environment for AI-enhanced engineering.