Earlier today, Brave Search announced it now has its own index for searching and finding images and videos on the internet. It's a major leap forward for the company in its quest to offer an independent and private alternative to search engines.
To get more info on this new index, we sent some questions to Brave Search via email. We received some answers from Arjaldo Karaj, the Director of Product at Brave Search.
First, was it always the goal of Brave to completely free itself from Microsoft and Google's search APIs or did this evolve over time?
Total independence was always the primary goal for Brave. When Brave Search launched in June 2021, about 13% of the queries required the help of third parties to achieve the desired level of quality across various types of queries. In less than a year after the initial release, we reduced that to only 7%.
This was made possible by the rapid adoption of Brave Search, as well as the broad adoption of our Web Discovery Project (which allows users to opt in and anonymously contribute browsing data to grow the Brave Search index). This enabled us to serve results that have the nuance and completeness expected by users, while addressing privacy and quality requirements.
These factors, coupled with the uncertainty over the future of the Bing API (which grew after the Microsoft and OpenAI partnership and the subsequent Bing API price increases), accelerated our plans to achieve 100% independence.
How hard was it to create your own API that would work as well as the ones from the large companies?
Brave Search is the fastest growing search engine since Bing, serving over 21 million queries per day. An important part of the quality of results in Brave Search is the quality of our index - one that is not as extensive as Google's, but one that definitely does not have all the noise and spam.
The newly released Brave Search API makes available many of these search capabilities in a simple, convenient way to customers of all scales, from whom we are getting great feedback.
What were some of the more surprising or unexpected things you encountered while developing the Brace Search index?
The web is full of noise. Filtering noise is a much harder problem than finding good pages.
The Web Discovery Project has had a crucial role. Users can opt into the WDP and anonymously contribute browsing data to grow the Brave Search index. This contributes to results having the nuance and completeness needed to compete on both privacy and quality.
Are you confident that Brave Search will be able to provide search results that people need when they try to find information on the internet?
Our focus is not on indexing every single page, but rather on focusing on the relevant pages and removing the noise. Brave Search is the default search engine in the Brave browser for users in the US, Canada, India, Germany, France, United Kingdom, Spain, Austria, Mexico, Brazil, and Argentina. It is also accessible to anyone at search.brave.com. While there are countries and languages we currently do better at, many users across the world have already chosen to use Brave Search and we are encouraged by the feedback we receive. We're aggressively assessing search quality across countries and languages, and we look forward to extending coverage.
How does Brave plan to monetize its Search Index so it can continue development?
Brave Search is ad-supported, but unlike other search providers, our ads adhere to the principle of privacy-first.
Users who want to support the Brave Search mission and get an ad-free experience can upgrade to Brave Search Premium.
Also, on May 30th this year, Brave launched the Brave Search API.
What can you tell us about Brave's privacy and security features now that it has its own Search index?
As always, our results preserve user privacy. Brave Search doesn’t track users, their searches, or their clicks. It’s impossible for Brave to share or sell user data because we don’t collect it in the first place; there is no concept of sessions either. And we’re always rolling out new features across our various tools to protect user privacy and put them in charge of their browsing experience.
Google and Microsoft are both jumping into generative AI as their next plan to offer search features. What do you think of this trend and does Brave have its own AI search plans?
This past March, Brave Search was the first to release a publicly available, search-integrated use of LLMs with a new feature called the Summarizer.
In response to the user’s input, and solely based on Web search results, the Summarizer provides concise and to-the-point answers at the top of Brave Search results pages. We trained our large language models (LLMs) to process multiple sources of information present on the Web. This is unlike a purely generative AI model, which is prone to spout unsubstantiated assertions, and thus the Summarizer produces a concise, accurate answer, in coherent language. Also, the provenance of original sources of data is cited at all times via links, so the user can easily check the sources or look for more details.
We are currently exploring how to apply LLMs not only to the search field, but also to the Brave browser. We think the assistant-like capabilities of LLMs will be extremely useful and will contribute to the user being in charge of their browsing experience.
What other new features is Brave working on for its browser and its Search index?
We have several developments in the works and will be making announcements in the near future.
Is Brave confident that it will be able to compete and even take some of the market share away from other browsers and search engines, particularly from Google?
Brave empowers users to browse privately, search privately, and offers a real alternative to Big Tech. Because of that, nearly 60 million users have already chosen Brave.
Our focus is on what is the next important step to take, or feature to build, that would add the most value to users.
Finally, is there anything else you wish to say about Brave's search index's new video and image search features?
We’re excited to offer our users a real alternative to Big Tech for their image and video queries. We know they value privacy-preserving image search and they were eager to have this available. We don’t have billions of dollars to spend on new features the way Big Tech does, but nevertheless our team’s creativity and focus enabled us to develop this in just a few months and attain 100% independence for all search results.
We would like to thank Arjaldo Karaj for taking the time to answer our questions.
3 Comments - Add comment