How does Ona work and what does it do?

Ona spins up fully configured development environments where AI agents can perform tasks. It allows for parallel work, prototyping, and inquiries directly within the development environment, aiming to boost developer productivity.

What are the key benefits of using Ona?

Ona offers parallelization of tasks, an agent that can co-author code, seamless integration with IDEs (browser and desktop), and fully sandboxed environments. This allows developers to work more efficiently and with less friction.

What are the challenges in building and using coding agents like Ona?

Challenges include user experience design (managing conversations and environments), making the technology accessible beyond expert users, and handling complex tasks like file edits. Users also face a learning curve in articulating specific prompts and decomposing tasks.

How does Ona handle different types of workloads (frontend vs. backend)?

Ona is designed to run anything that runs on Linux, making it capable of handling both frontend and backend workloads, including complex infrastructure and backend services, unlike some solutions focused solely on web containers.

What is Ona's pricing model?

For enterprises, Ona offers flexible pricing and can be deployed within the customer's VPC. For self-serve users, there's a credit model combining compute and token usage, with a $100 credit offer for upgrading to the core tier in September.

Will Ona eventually replace traditional IDEs?

While Ona integrates with existing IDEs and allows users to 'pop out' to their preferred editor, the shift is towards new interfaces that support parallel work. The true change may come from evolving workflows and cognitive demands rather than just compute location.

How does Ona ensure security with AI agents having broad access?

By default, Ona environments are fully sandboxed and isolated, meaning agents don't run on your local machine. While agents have broad access by default ('YOLO mode'), enterprise customers can configure deny lists for specific commands or actions.

⚡️Launching Ona: Coding Agent with Fully Sandboxed Cloud Environment

Latent Space Podcast

Science & Technology5 min read34 min video

Sep 1, 2025|2,404 views|47|9

Save to Pod

Key Moments

TL;DR

Ona launches as an AI coding agent with sandboxed cloud environments, evolving from Gitpod.

Key Insights

Ona is a new AI-powered software development platform from the creators of Gitpod, designed for the future of AI-driven development.

The platform offers fully configured, sandboxed cloud development environments that can run any Linux-based workload, integrating with existing enterprise infrastructure.

Ona enables developers to run complex back-end tasks, manage prototypes, and perform tasks in parallel without altering environment state, significantly boosting productivity.

The transition from Gitpod to Ona reflects a strategic pivot towards embracing AI coding agents and outgrowing the previous technical name and architecture.

Ona's design prioritizes user experience and accessibility, aiming to make advanced AI development tools usable for a broader range of developers, not just experts.

The platform supports a flexible pricing model, offering credits for self-serve users and custom solutions for enterprises, with a focus on value provided by isolation and autonomy, not inference.

THE EVOLUTION FROM GITPOD TO ONA

The creators of Gitpod have launched Ona, a new platform that marks a significant evolution from their previous work in cloud development environments. Initially focused on providing a one-click, browser-based IDE experience with Gitpod, the company has recognized the transformative potential of AI coding agents. This realization, particularly accelerated by advancements in models like GPT-4, led to the strategic decision to pivot and rebrand as Ona. The new name and focus reflect their commitment to building a platform specifically designed for the burgeoning era of AI-driven software engineering, moving beyond the technical connotations of 'Git' and 'Pod'.

CORE FUNCTIONALITY AND SANDBOXED ENVIRONMENTS

Ona's core offering is a sophisticated AI coding agent operating within fully sandboxed cloud environments. These environments are capable of running any workload that functions on Linux, providing a secure and isolated space for development. This deep integration allows Ona to connect with enterprise-specific resources like on-premise databases, registries, and secret managers, enabling software development within the company's Virtual Private Cloud (VPC). The platform emphasizes reproducibility and standardization, ensuring that developers and AI agents work within consistent, pre-configured setups.

ENHANCING DEVELOPER PRODUCTIVITY WITH PARALLELISM

A key advantage of Ona is its ability to significantly enhance developer productivity through parallelism and autonomy. Developers can use Ona to handle complex back-end tasks, prototype new features, and run inquiries or development tasks concurrently without interfering with each other or the main development state. This allows for a more dynamic and efficient workflow, where multiple activities can be managed simultaneously. The platform ensures that these parallel operations do not lead to environment state corruption, maintaining a clean and consistent workspace for all tasks.

DEMONSTRATION OF ONA'S CAPABILITIES

A live demonstration showcased Ona's practical applications, from simple inquiries about codebases to complex prototyping and debugging. Users can interact with Ona conversationally, asking it to examine code, build prototypes, or investigate potential issues within the development environment. The platform seamlessly integrates with developer workflows, allowing for immediate access to a VS Code instance within the browser or a seamless transition to local desktop IDEs if needed. The ability to track changes, manage different conversations in parallel, and maintain environment state makes Ona a powerful tool for day-to-day coding tasks.

ADDRESSING DESIGN CHALLENGES AND USER EXPERIENCE CHALLENGES

The development of Ona involved overcoming significant design challenges, particularly around user experience and agent accessibility. A primary hurdle was determining the optimal mapping between development environments and agent conversations, and making these powerful tools approachable for a wide range of developers, including those who may be AI-skeptical. The platform aims to bridge the gap between basic autocomplete features and the potential of advanced AI, ensuring that the interface and interactions are intuitive and not overly complex, which is crucial for broader adoption.

TECHNOLOGY STACK AND ARCHITECTURAL DECISIONS

Ona is built on a foundation of robust infrastructure, including its own orchestration system and a strong emphasis on reproducible environments defined by dev container configurations and automation YAML files. Unlike some competitors focusing on web containers or front-end workloads, Ona supports any Linux-based application, including complex back-end services written in Go, Python, and TypeScript. The platform currently leverages models like Sonnet 4 but is designed to be flexible with different LLMs. A critical aspect of its development has been refining the file editing capabilities of agents, which has proven to be a surprisingly complex yet vital component.

PRICING MODEL AND BUSINESS STRATEGY

Ona employs a flexible pricing model tailored for both enterprise clients and self-serve users. For enterprises, the platform offers deployment within their VPC, providing enhanced security and flexibility, with pricing arrangements that reflect the value of specialized integration. The self-serve tier, launching with a credit model, blends compute and token usage. The company emphasizes that their pricing is based on the value derived from Ona's unique features—sandboxed isolation, connectivity, parallelism, and advanced editing capabilities—rather than solely on inference costs. They are not building their own LLMs but betting on frontier models focused on coding.

THE FUTURE OF IDEs AND DEVELOPER WORKFLOWS

Ona's trajectory suggests a future where traditional IDEs may evolve or be complemented by new interfaces designed for parallel work and agent collaboration. While acknowledging the deep-seated habits and preferences for local IDEs, the platform advocates for interfaces that can manage multiple tasks simultaneously without overwhelming the developer. This shift is driven by the increasing autonomy of AI agents. Ona aims to provide developers with choices, supporting conversational interfaces, browser-based IDEs, and traditional desktop IDEs, prioritizing the right tool for the task at hand as workflows adapt to AI integration.

EMPHASIS ON SANDBOXING AND AUTONOMY OF AGENTS

A core tenet of Ona is the secure and autonomous operation of AI agents within isolated environments. By default, environments are highly permissive, allowing agents significant freedom to perform tasks. This 'YOLO mode' is a deliberate design choice to maximize productivity, with the understanding that accidental mistakes or explorations are risk-free due to the ephemeral nature of these environments. For enterprises with stricter compliance needs, a deny-list feature can be implemented to restrict specific agent actions, ensuring a balance between agent autonomy and controlled execution.

MOBILITY AND ACCESSIBILITY OF DEVELOPMENT TOOLS

Ona's design extends to providing a powerful development experience across various devices, including mobile and tablets. The platform's ability to maintain identical environments across different devices ensures a seamless transition for developers. This allows for increased productivity even when using mobile devices, enabling tasks that were previously confined to powerful workstations. The browser-based VS Code and overall environmental consistency make it possible for developers to work effectively on the go, fitting development into various aspects of their lives, even during unconventional hours or while commuting.

Mentioned in This Episode

●Products

●Software & Apps

●Companies

●Organizations

Ona: Your Agentic Development Environment

Practical takeaways from this episode

Do This

Start small with AI agent tasks and gradually increase complexity.

Be excessively specific in your prompts to the agent.

Leverage pre-configured environments for compliance and standardization.

Utilize VS Code in the browser for quick edits and reviews.

Seamlessly transition to desktop IDEs (like VS Code, JetBrains) when needed.

Explore the mobile experience for productivity on the go.

Give agents autonomy in sandboxed environments, mistakes are risk-free.

Use the SDK to integrate Ona into your custom workflows.

Experiment with running multiple agents in parallel.

Avoid This

Make assumptions about what the agent knows or can do.

Give vaguely specified or overly large tasks to the agent.

Rely solely on Docker for security boundaries.

Worry about dangerous permission flags when using sandboxed environments.

Hesitate to throw away ephemeral environments after use.

Avoid using your race car (highly focused IDE) for shopping (simple tasks).

Common Questions

Ona is a new coding agent and platform built by the GitPod team. The rebrand from GitPod to Ona signifies a shift in focus towards agentic software development, with Ona being a core part of the user experience and platform rather than just an agent.