Transforming AI Search Integration: Perplexity’s Carbon Partnership

In 2024, Perplexity stands out as a remarkable force in the AI search integration arena. Established by Aravind Srinivas, a former researcher at DeepMind and OpenAI, this startup has successfully attracted hundreds of millions in funding, boosting its valuation to a staggering $9 billion. Over the past year, Perplexity has introduced a host of innovative features, including Pages, Spaces, and dynamic shopping options, which have strengthened its position in the market.

These pioneering advancements position Perplexity as an “AI-first” knowledge discovery platform, setting it apart from conventional search engines like Google and Bing that enhance their existing systems with AI functions. However, this is just the beginning for Perplexity as it navigates increasing competition.

Bridging the Data Gap Through Carbon Integration

To amplify its services, Perplexity has taken a significant step by acquiring Carbon, a startup focused on tackling the “data gap” many businesses encounter in the realm of AI search integration. This acquisition aims to streamline the knowledge discovery process within various workflows significantly.

Carbon has developed a powerful retrieval framework that enables seamless connections between external data sources and large language models (LLMs). Users can leverage the Carbon universal API or SDKs to synchronize their data sources and retrieve relevant information efficiently. Carbon’s system is compatible with over 20 data connectors and accommodates more than 20 file formats, including text, audio, and video files.

The Growing Landscape of AI Search Integration

Today, AI search integration plays a pivotal role in business workflows for diverse users, from individuals to entire teams. This technology empowers users to receive immediate answers to their queries without wading through countless links or content pieces. Perplexity has thrived by adopting this streamlined approach, utilizing a mix of large language models to gather information swiftly and effectively.

Furthermore, Perplexity enables teams to extract valuable insights from personal or organizational files, including PDFs and Word documents. However, a critical challenge lingers. With the internet housing a vast pool of public information, businesses frequently grapple with huge amounts of proprietary data. This situation can make it quite impractical for users to upload internal files—like PDFs and other documents—individually. As a result, the quality of search outcomes may decline, leading to generic responses devoid of crucial context relevant to the organization.

Industry experts, such as Sanjeev Mohan, previous VP of Research at Gartner for data and analytics, predict a surge in the extraction and transformation of unstructured data by 2025. This anticipated trend will empower teams to utilize their dispersed internal data sources more effectively, ultimately enhancing LLMs’ ability to deliver precise and contextualized answers.

The Value Carbon Adds to Perplexity

With the integration of Carbon, Perplexity’s search platform will gain advanced capabilities through the incorporation of various data connectors. This integration will facilitate seamless connections for users between their data sources, such as Google Docs, Notion, HubSpot, and Slack, directly within the Perplexity framework.

• For instance, if a user needs to update a product launch date and is looking for the most recent deadline set by the team, Perplexity can quickly analyze all relevant data across these various platforms.
• Essentially, users won’t need to manually search for context across different applications or the web; the platform will autonomously present the information they need.

Sara Platnick, who heads communications at Perplexity, emphasizes that this setup enables the technology to find answers without requiring users to pinpoint the exact document or database containing the information. This feature significantly enhances user experience and operational efficiency.

Moreover, another potential use case involves extracting insights from customer meetings. Perplexity could rapidly fetch conversation details and insights from connected CRMs, saving users precious time.

By adopting Carbon’s retrieval-augmented generation (RAG) workflows, Perplexity aims to make enterprise search more user-friendly. This strategy alleviates the difficulties companies often encounter when attempting to build their own RAG pipelines from zero.

• Kevin Petrie, VP of Research at BARC US, notes that early adopters of this technology primarily concentrate on:
• Customer service
• Document processing
• Image processing
• Recommendation engines

The Significance of Effective Execution

Though acquiring Carbon signifies a notable advancement for Perplexity, the real challenge lies in seamlessly integrating Carbon’s technology. A smooth and secure execution is crucial, particularly when managing proprietary data from essential knowledge repositories within businesses.

Many firms remain wary about exposing their intellectual property to outside platforms. Thus, Perplexity and Carbon must establish stringent governance controls to ensure that companies can keep their data’s privacy and security protected within their own frameworks.

Sara Platnick reassures users that all data from internal and private sources will be encrypted while stored in Perplexity’s system. In addition, protective measures will be in place to prevent unauthorized access to private documents, ensuring that robust security protocols are implemented.

Currently, there is no set timeline for integrating Carbon into the Perplexity platform. However, users should be aware that Carbon plans to cease operations of its managed API on March 31, 2025. Existing API clients have been informed of the transition, and the Carbon team is prepared to assist them during this process.