In order to produce high-quality code that is suitable for enterprise usage, a thorough code review is an essential step in the software development process. But when thousands of lines of code need to be reviewed by professionals to find bugs or other problems, it may be a time-consuming procedure. Because people are human, traditional reviews may be arbitrary and inconsistent. Numerous new AI-powered code review tools have emerged in the market as generative AI is increasingly incorporated into the software development process. These tools help software development teams improve quality and produce clean code more quickly, while also increasing developer productivity and job satisfaction.
An inventive firm called Bito makes AI agents for a variety of software developers. When it introduced its first AI-powered developer agents in 2023, introducing large language models (LLMs) to code-writing environments, it became a leader in its area. With its growing portfolio of AI-powered agents, Bito’s executive team aims to unite developers and AI in powerful new ways, viewing them as essential components of the future. Pull request (PR) time-to-merge can be accelerated by up to 89% with its flagship product, AI Code Review Agent, which speeds up development timelines and frees up developers to concentrate on creating code and innovation. Bito’s AI Code Review Agent is designed to support senior software engineers’ review, not to replace it. It does this by organizing, summarizing, and then suggesting line-level modifications with code base context, allowing engineers to concentrate on more strategic parts of code review, including assessing business logic.
With over 100,000 developers using its products worldwide, Bito has demonstrated success in streamlining code review through generative AI, providing remarkable outcomes and business value to its clients, which include Gainsight, Privado, PubMatic, On-Board Data Systems (OBDS), and top software companies. Customers of Bito have discovered that AI Code Review Agent offers the following advantages:
- Speeds up PR mergers by 89%.
- 34% fewer regressions
- Provides 87% of the reviewable feedback required for a PR.
- Has a signal-to-noise ratio of 2.33.
- Is compatible with more than 50 programming languages.
Despite the impressive results, Bito had to overcome developers’ natural skepticism when assessing a plethora of new AI-powered tools. In order to expedite and broaden the adoption of AI Code Review Agent, Bito executives intended to introduce a free tier option that would allow potential customers to test out the features and benefits of AI Code Review Agent and entice them to upgrade to the full, paid Teams Plan. Its Teams Plan would offer more sophisticated capabilities like downstream impact analysis and one-click acceptance of line-level code proposals, while its Free Plan would offer AI-generated PR summaries to give a summary of changes.
In this article, we describe how Bito uses Nova to provide a free tier option for its AI-powered code evaluations.
Choosing A Cost-Effective Model For A Free Tier Product
Bito wanted a foundation model (FM) that would deliver the appropriate performance and outcomes at a fair price in order to offer a free tier option for AI Code Review Agent. Naturally, Bito would have to pay inference expenses if it were to offer code reviews to prospective clients for free. Bito evaluated a wide selection of models, including the top-performing FMs on Bedrock and the OpenAI GPT-4o small, over the course of two weeks in order to choose one for its Free Plan. The team was especially interested in the Nova models, which are quick and affordable versions that were just released on Bedrock.
Bito concluded that Nova Lite provided the ideal balance of cost-effectiveness and performance for its use cases. Because of its speed, code review summaries may be created quickly. Cost, a crucial element of Bito’s Free Plan, turned out to be the decisive element, nonetheless. In the end, Bito’s requirements for speed, price, and quality were satisfied by Nova Lite. Bedrock and Nova Lite work together to enable Bito to provide the dependability and security that their clients required when they trusted Bito with their code. After all, one of Bito’s main pledges to its clients is meticulous code management. Code is neither stored nor used for model training. Additionally, its products are SOC 2 Type 2-certified to offer confidentiality, privacy, processing integrity, and data security.
Selecting The Right Models For Different Product Tiers
Bedrock is now Bito’s standardized platform for model exploration, addition, and execution. The unified Bedrock API and controls are used to access and integrate Bito’s Free Plan, which is based on Nova Lite, and Anthropic’s Teams Plan, which is for-profit and powered by Claude 3.7 Sonnet. When consumers upgrade from Nova Lite to Anthropic’s Sonnet, Bedrock allows for a smooth transition with few code modifications. Bito executives are careful to note that Nova Lite inspired their Free Plan, not only powered it. It would not have been able to offer a free tier of AI Code Review Agent without the extremely low cost of Nova Lite, which they saw as a calculated move that would allow them to grow their enterprise clientele. Three times as many potential consumers were drawn to its Free Plan as expected as a result of this strategy’s rapid outcomes. Many people upgrade to the complete AI Code Review Agent once the 14-day trial period is over in order to take use of all of its features.
Driven by the success of AI Code Review Agent, Bito is now leveraging Nova Lite to power the chat features of its latest AI agentic technology, Bito Wingman, a full-featured developer assistant in the integrated development environment (IDE) that includes architectural advice, code generation, and error handling, among other things. Once more, Bito chose Nova Lite because of its affordable price and high quality.
In Conclusion
In this piece, we explained how Bito, a cutting-edge firm with an expanding line of AI-powered developer agents, decided to use Nova Lite to power its main product, AI Code Review Agent, which is available for free. Its AI-powered agents are made especially to ease developers’ burdens and increase the impact of their work:
- Bito was able to overcome one of its main business obstacles—luring in enterprise clients—thanks to Nova Lite. Bito’s generative AI-driven flagship product, AI Code Review Agent, garnered three times as many potential new customers after launching a free tier.
- During extensive internal testing, Nova Lite performed better than other models, offering Bito the performance it required at the extremely low price point required to introduce a free tier of AI Code Review Agent.
- Bedrock gives Bito the ability to easily switch between models for each AI Code Review Agent tier, including Anthropic’s Claude 3.7 Sonnet for its for-profit Teams Plan and Nova Lite for its Free Plan. Additionally, Bedrock offered security and anonymity, two important factors for Bito clients.
- Bito demonstrates how forward-thinking companies may use Nova Lite’s quality, affordability, and timeliness to provide value to both their clients and their company.
Amar Goel, co-founder and CEO of Bito, says, “Our challenge is to push the capabilities of AI to deliver new value to developers, but at a reasonable cost.” “Nova Lite provides us with the incredibly quick and affordable model we require to support our AI Code Review Agent’s free offering—and draw in new clients.”