"The primary reason we went with Ona (formerly Gitpod) was onboarding.
Easier onboarding for developer environments and faster time to first commit. Ona pushes that value proposition further because, one of the hardest things as an engineer onboarding to a new project and a new repository is understanding what was there before.
Ona helps by having an assistant there, who you can ask those questions of (even if you're not having it write code). Ona takes what was already a great value proposition and drives it even further. Faster startup time, faster onboarding time.
It all pairs together in a really good way."
- Patrick, Kingland
—
Kingland, a leading enterprise data management provider has transformed their development workflow using AI agents, and automated and secure development environments. Their use cases include dependency updates, code refactors and test improvements. Using Ona, Kingland ensures their sensitive financial services code and data remains secure utilizing private LLM integration via AWS Bedrock, while also enabling developers to run multiple agent tasks securely in parallel. Patrick serves as Enterprise Architect at Kingland, where he leads engineering and research with responsibility for developer productivity.
Key outcomes:
- 10x faster migrations - Jest v30 migration completed in 30 minutes vs 5 hours
- 15-year legacy codebase - Documented with agent generated JavaDocs
- Parallel AI execution - Transforms downtime between meetings into productivity
- Not only for developers - Tech writers and data scientists using Ona for docs and SQL
- Full compliance - Maintained while enabling autonomous AI agents
Background
Kingland's journey to AI powered development started with Windows machines distributed manually as VMware images. Kingland improved on this setup initially by migrating to HashiCorp Vagrant with centralized VM images, however performance issues persisted and engineers with 32GB laptops continued to face slow environments.
The breakthrough came with Ona (formerly Gitpod), through automated and standardized environments running in the cloud where Kingland eliminated thousands of hours lost on environment setup issues. As Patrick recounts: Developers were ‘spending tons of hours per year fixing development environments,’ often losing ‘upwards of two days setting up their environment when something went wrong.’
With their ephemeral development environment platform in place, Kingland was uniquely positioned to adopt agents into their workflows. The combination of ephemeral workspaces and AI agents created a powerful combination, granting agents greater autonomy through access to a secure, on-demand isolated environment "you don't care if it [the agent] wipes out your Ona environment because it’s disposable, but you do care if it wipes out your whole computer."
Kingland’s use cases for Ona
Kingland has discovered three use cases where Ona delivers value:
- Dependency management - The most dramatic success came with automated dependency updates. Kingland uses Renovate to manage dependencies where Patrick noted that a Jest v30 migration was completed with "about 15 to 30 minutes of actual engineering interaction" compared to "what probably would have taken an engineer maybe four to five hours." Engineers now simply provide Ona with migration notes and return later to review completed tested and migrated code.
- Code refactoring - Refactoring represents another high-value use case. Patrick highlighted two approaches: "you ask it to either refactor test files without changing production files, or refactor production files without changing test files. Both of those use cases are really good because you can give it instructions and you have a safeguard for whether it did what it was supposed to do correctly."
- Quality improvements - Legacy code documentation emerged as a success story. When Kingland implemented CheckStyle requiring JavaDoc comments on all methods in their 15-year-old Java codebase, teams faced the daunting task of documenting reams of previously untouched code. Patrick described how engineers would delegate the refactoring work to Ona: "when we're working on a package, we’d just ask Ona to go and write comments for all of the methods on that package. We’d give it examples of good commands and bad comments, then let it run for a while and review the comments it created."
Secure deployment of Ona
Kingland's adoption of Ona demonstrates how AI agents can meet the stringent security requirements of regulated industries meeting compliance requirements. For this Kingland needed an AI solution that wouldn't compromise their security posture.
Key security features of Ona:
- Amazon Bedrock LLM integration - All LLM processing happens through Kingland's own AWS account. As Patrick noted: "AWS Bedrock models have a much stricter security stance than what you see from model providers directly." This was crucial for Kingland, who already uses Bedrock for AI features in their own products. The integration ensures they maintain full control over their data with no sharing of IP with third parties, and no model training on their code.
- Ephemeral environment security - Each AI agent runs in an isolated, disposable environment that's destroyed after use. Patrick emphasized: "The best way to handle security is to not have environments that are alive for long enough to get compromised. If it's ephemeral, just destroy it and start over from scratch." This approach eliminates persistent attack surfaces and ensures that even if an AI agent were to be compromised, the blast radius is limited to a single, temporary environment.
- Self-hosted environments - Kingland deployed a dedicated runner enabled with AI, allowing Kingland to control who has access to the feature. The architecture gives organizations like Kingland the means to precisely control which teams and projects have access to agents by controlling access to who has access to the runner.
- Agent and user audit trails - Critical for regulated industries, Kingland can now maintain a complete log of who initiates environment actions and what changes are made. Patrick highlighted this as "so critical for us to keep track of the audit trail of who performs a given action."
- Network security - Integration with Kingland's existing VPN and IP allowlisting provides additional layers of protection. The enterprise runner operates within their security perimeter, with load balancers configured to accept connections only from approved IP ranges, ensuring AI capabilities are accessible only from within their corporate network.
Kingland's journey with Ona demonstrates that enterprises don't have to choose between driving productivity with agents and security. By building on their installed infrastructure for standardized development environments orchestrated by Ona, Kingland have enabled their development teams to accelerate their output securely.