With WarpBuild, our objective is to create a set of products that supercharge developer productivity across the entire software development lifecycle.
WarpBuild CI offers fast, reliable, and scalable build infrastructure for all your code. I am excited to announce the second product in our suite: Helios, the Code Intelligence Platform.
What is Helios?
Helios is a code intelligence platform that works where you work and adapts to your workflows. The first feature of the Helios suite is a code review agent, that operates with complete context of your codebase.
The Helios code review bot can be added to your GitHub organization, and it will automatically review your pull requests for the repositories to which you provide access.
What does it do?
Here are the things that Helios does:
- Adds no-noise recommendations on pull requests.
- Adds a summary of the changes in the pull request.
- Provides a file-by-file walkthrough of the changes in the pull request.
- A rough estimate of the effort required to review the pull request so you can prioritize your reviews.
Equally importantly, there are configuration options to customize the bot's behavior to your liking. These preferences can be set at an organization level and inherited by all repositories, with an option to override at a repository level.
- Set trigger conditions for when the bot should review a pull request.
- Trigger on every pull request.
- Trigger only when there are certain labels on the pull request.
- Trigger only on certain branches.
- Trigger reviews manually via the dashboard.
- Provide custom instructions to the bot at a per-repository level.
- Configure ignore patterns for files that must not be indexed.
The roadmap here is mounting, and we want to hear from you to help us prioritize the features. Please reach out to me on email or X if there is anything you'd like to see.
Should you use Helios?
We have been internally using Helios for a while now, and it has been a great tool for us. It has helped us catch a lot of bugs and improve the overall code quality of our codebase.
I personally rely fairly heavily on Helios for summaries and walkthroughs. It also catches bugs that are hard to catch otherwise.
How we built it
We took an early decision to build the entire backend in TypeScript. Type safety is a non-negotiable for us. Another reason is that, as we go into an increasingly LLM-centric world, types and linting are more important than ever in getting the most out of LLMs. This is crucial in enabling a small team to build great products.
Here are the different components that enable Helios:
- Agni: our custom agent framework
- Hatchet: a postgres-based durable queue for running agents
- Langfuse: for tracing and monitoring agents
- Postgres + pgvector: the backend and vector DB for storing and querying embeddings
A strong factor for us is to be able to deploy the product on any infrastructure and not be vendor locked. This is important as we take on larger customers.
A longer blog post on the technical details of the product is coming soon.
Agni: our custom agent framework
Our ambitions for Helios are huge. Our vision for this encompasses a lot of different surfaces, modalities, workflows, and user experiences. Most older frameworks are not well suited for this. Some promising frameworks like crew.ai are python-centric.
We leverage Vercel AI SDK to abstract the underlying LLM provider as a very thin wrapper around the LLM providers.
Agni has high-level abstractions for flows, agents, tasks, and tools. It is a graph based framework that easily allows expression of complex workflows, or individual agent invocations. I intend to write a detailed blog post on this in the future.
While this was a significant time investment, we believe this is a good long-term bet.
Data Security
We take data security and privacy very seriously. We are SOC 2 compliant and have a robust data security infrastructure in place. Each product within WarpBuild (CI Runners, Helios) have their own isolated GitHub integrations each with their own permissions, and separate data stores.
Helios takes access to your codebase and metadata from your GitHub account. This is required to provide the code intelligence features. However, we do not use this codebase for any purpose other than to provide the code intelligence features to you. We do not train models or share this data with any third parties. We work only with LLM providers that have strong security and privacy guarantees, and have Zero Data Retention agreements with them.
Try it out
You can try out Helios for free. Sign up here. Teams get a 14 day free trial.
Founder Note: Why Helios
WarpBuild was formed to improve developer productivity across the entire software development lifecycle.
CI speeds are a known bottleneck and a major source of friction. Over the last year, we built a set of features to address this. We are continuing to invest significantly in this area, including some very exciting features that we will be announcing soon.
The broader industry is now at a point where it is possible to build a set of products that address many of the remaining gaps in the software development lifecycle as well. The emergence of reasoning LLMs, affordability, performance and quality improvements especially for code make this possible.
It is not a question of if, but when we will see autonomous agents that understand code in the context of the organization and can help developers with their day to day tasks.
We are excited about the future of agents that can reason about the codebase in the context of the organization, adapt to your workflows, and can prioritize, collaborate, design, and engineer the product with you, leaving the implementation to code gen tools like Cursor, Windsurf, and others.
Helios code reviews is the first step in this direction.
Note
Try out Helios for your code reviews with a 14 day free trial. Get started today.