As AI search gradually becomes a consensus among everyone, the threshold for entrepreneurship is also getting higher and higher.

Yesterday, according to Techcrunch, the AI search engine Genspark, founded by two former executives of Baidu, announced the completion of a $60 million seed round of financing, with a valuation of $260 million, led by BlueRun Ventures.

Genspark's product features are quite interesting. When most AI searches struggle with information illusions and lack of solutions, it focuses on high content accuracy.

When users ask questions, Genspark can not only provide answers to users but also generate a single-page overview similar to Baidu Baike based on the collected information, called Sparkpage.

In Genspark's vision, Sparkpage is not only a revolutionary new form of webpage but also an important way to change the poor quality of existing web content. So far, according to the Genspark official website, users have created and generated 200,000 Sparkpages.

The design of Sparkpage will determine whether Genspark can ultimately break through in the AI search competition.

/ 01 / Sparkpage, the AI version of "Baidu Baike"

Compared to other AI searches, the biggest difference of Genspark is that it not only provides answers but also generates Sparkpages based on the search content. This is a single-page overview composed of websites and web content, similar to Baidu Baike.

For example, when I ask Genspark, "Who is Raven Intelligence?" Genspark will quickly provide basic information about Raven Intelligence, like this:

Below the answer, Genspark has also created two Sparkpages related to Raven Intelligence, with roughly the same content, including basic introduction, publishing platforms, areas of focus, and representative articles, etc.

On the right side of the Sparkpage, there will also be an AI assistant to help users with problem extension and information expansion.

Why does Genspark want to create something similar to Baidu Baike? This stems from the thinking of Genspark's management. They observed that the internet is full of various pieces of junk information, and people often find it difficult to locate high-quality information.

Therefore, Genspark not only wants to help users find answers faster through AI search but also hopes to establish a high-quality content unit in this process, which is now Sparkpage.

To achieve this, Genspark uses three methods:

First, let the AI model adopt more authoritative and popular webpages to enhance the depth and breadth of usable information. Genspark's founder Jing Kun said that the company plans to purchase copyrighted content in appropriate situations to improve the accuracy of the engine.

Second, utilize AI technology to periodically cross-check this information with trusted databases and authoritative sources;

Third, ensure that Sparkpage content is not influenced by commercial influence or editorial bias.

For example, Genspark will support user-initiated modifications and editing to allow users to verify claims. Additionally, Genspark's AI system will improve future results based on each edit.

Focusing on information authenticity is a very differentiated idea. Because since the emergence of AI search, information illusions have always been the most criticized aspect of such products, such as Google's AI suggesting applying glue to pizza or Arc Search's claim that severed toes will eventually grow back.

Currently, there is no evidence that Genspark can ensure that search content will not have illusions or other negative content. But at least it can be determined that the existence of Sparkpage gives Genspark more possibilities.

/ 02 / The possibility of building a content ecosystem