Neeva CEO Sridhar Ramaswamy on AI chatbots and the search revolution | E1686

TL;DR

Neeva combines its internet-scale search system with language models to deliver direct, reliable, cited, and timely answers while prioritizing privacy and an ad-free experience. CEO and co-founder Sridhar Ramaswamy says the 50-person team can generate fluid answers for more than half of submitted queries and expects coverage to reach 75–80% of suitable queries. Read on to understand Neeva’s technology, accuracy strategy, business model, and vision for answer-based search.

Transcript

hey everybody happy Monday Jason is still away if you follow him on Instagram you can see his ski adventures in Japan it's incredible I am back though today with another great interview today on the show I'm joined by sridhar ramaswamy the CEO and co-founder of Neva an ai-powered search engine and chatbot it's private it's AI powered search but it'... Read More

Key Insights

• ✊ Neva aims to revolutionize search by providing concise and reliable answers through AI-powered technology.
• 👻 The combination of a search engine and language models allows Neva to generate authoritative and timely responses.
• 👤 The company prioritizes user experience, privacy, and all legal content without content moderation.
• 🫡 Neva compensates publishers and explores partnerships to display their content while respecting copyright and fair use policies.
• 🥺 The future of search is shifting towards answer-based paradigms, leading to potential disruptions in the publishing industry.
• ✊ Content moderation and filtering present challenges and potential legal implications in the AI-powered search landscape.
• 👨‍🔬 Neva's approach to search emphasizes competition, user control, and personalization to enhance the overall search experience.

Questions & Answers

Q: What are Neeva and Neeva AI?

Neeva is a next-generation search engine led by CEO and co-founder Sridhar Ramaswamy. It integrates language models with a search system to provide direct, reliable, cited, and timely answers while emphasizing privacy and an ad-free experience.

Q: How does Neeva differ from ChatGPT and Google Bard?

Neeva runs a search engine and language models in parallel instead of relying exclusively on open-loop generation. Its goal is to preserve search qualities such as quick, authoritative, and timely information while using language models to produce fluid answers.

Q: How does Neeva make its AI-generated answers more reliable?

Neeva constrains generation using content retrieved from trusted sources rather than letting the model respond only from its broad training. It supports answers with relevant citations so users can see the information behind a response.

Q: How many searches can Neeva answer directly?

Ramaswamy says Neeva’s 50-person team can generate answers and fluid responses for more than half of the queries entered into the service. He says the company expects to provide answers for 75–80% of suitable queries.

Q: Why did Sridhar Ramaswamy co-found Neeva?

After close to 16 years at Google, including work on the early search ads team and leadership of the ads and commerce teams, Ramaswamy wanted to reimagine search. He and co-founder Vivek believed Google was trapped by its own success and sought to create a product centered on serving the user.

Q: What search infrastructure has Neeva built?

Over its first three years, Neeva built a search stack, operated a crawl at internet scale, and developed its own search system. The company then began deeply integrating language models after seeing early versions of technologies such as GPT-3.

Q: How does Neeva handle privacy and advertising?

Neeva was designed with an early focus on privacy and an ad-free experience. Ramaswamy describes privacy as a feature of a strong user-focused product rather than the entire product by itself.

Q: How does Neeva make money without relying on ads?

Neeva uses a freemium model with a basic free product and a paid subscription option. It also works with publishers and shares a portion of subscription revenue with them.

Summary

In this interview, Sridhar Ramaswamy, the CEO and co-founder of Neva, an AI-powered search engine and chatbot, discusses the unique model of their search engine, the challenges they faced in building it, and the future of search. He highlights the importance of privacy, efficiency, and providing accurate and reliable answers to users. The conversation also touches upon the impact of Neva's disruptive model on publishers and the ad-supported publishing ecosystem.

Questions & Answers

Q: What is Neva and how does it differ from other AI-powered search engines?

Neva is an AI-powered search engine and chatbot that focuses on privacy and providing accurate answers to users. Unlike other search engines like OpenAI Chat GPT and Google's Bard, Neva takes a different approach by combining a search engine and large language models in parallel. The goal is to provide quick, authoritative, and timely information to users.

Q: How did Sridhar Ramaswamy come up with the idea for Neva?

While Sridhar was at Google, he realized that people prefer getting answers directly rather than clicking on links. He noticed the positive response to Google's featured snippets, where a short answer is provided on the search results page. This insight led him to believe that answers are more valuable than links and sparked the idea for Neva.

Q: What is the technological underpinning of Neva?

Neva utilizes large language learning models, similar to GPT-3, but combines them with a search engine. The search engine identifies authoritative sources and extracts relevant information, which is then summarized to generate concise answers. This multi-step process ensures the answers are accurate and reliable, avoiding the pitfalls of open-loop language models.

Q: How difficult was it to build Neva's search engine and integrate language models?

Building Neva's search engine was a challenging task, considering the scale at which it operates. The team had to overcome technical hurdles to ensure efficient retrieval and summarization of web pages. Additionally, integrating language models within the search stack required inventive techniques to make them smaller and more cost-effective. It was a combination of years of experience and experimentation that led to the current working model.

Q: How does Neva make money?

Neva follows a freemium model, where the basic product is free to use, and users have the option to subscribe to a paid version for additional benefits. The paid subscription includes unlimited searches, access to connectors for various accounts, and personalized features. Neva also partners with other companies to provide additional perks, like VPN services and password managers, to its paid subscribers.

Q: How does Neva position itself for growth and potential 100x investment?

Considering the vast potential market for search, even a small fraction of the global population becoming Neva subscribers would be enough to sustain the company. By estimating the number of subscribers required to run a search engine for the whole world, Neva can project substantial annual revenue. Additionally, Neva explores business-to-business opportunities by licensing its technology to other search providers or offering search APIs to platforms like Reddit.

Q: What are the challenges and consequences for publishers due to Neva's disruption in the search ecosystem?

Neva's focus on providing concise answers and reducing reliance on links can disrupt the traditional search engine and publisher model. Publishers heavily dependent on ad-supported search traffic may face challenges as users obtain answers directly from Neva's snippets. This disruption could fundamentally change the relationship between search engines and websites, forcing publishers to reconsider how they attract and retain customers. However, it also presents opportunities for publishers to offer conversational interfaces and more engaging experiences on their websites.

Q: Why is privacy an important aspect of Neva's search engine?

Privacy is a significant aspect of Neva, as users often search for queries that are deeply personal. Neva respects user privacy by not tracking search queries by default. Additionally, Neva offers features for personalized preferences, allowing users to set their preferred sources and control their search results. By prioritizing privacy, Neva aims to provide a more peaceful and focused search experience for its users.

Q: What feedback has Neva received from its users regarding their product experience?

Neva has received positive feedback from users who appreciate the privacy-focused and ad-free experience it provides. Users described using Neva as a breath of fresh air, making the online world feel quieter without constant tracking and distractions. Additionally, users find the short answer format to be efficient and engaging, particularly in an era of limited attention spans.

Q: How does Neva address the concern of filter bubbles and biased information?

Neva aims to give control back to the user by allowing them to express their preferences for preferred sources of information. Neva's algorithms consider authority, credibility, and different viewpoints when generating search results. Additionally, Neva offers features like "Bias Buster" that display viewpoints from different ends of the political spectrum, enabling users to explore a more diverse range of perspectives.

Takeaways

Neva's unique model, combining a search engine and large language models, allows for the efficient retrieval and summarization of information, resulting in concise answers. The focus on privacy, accuracy, and reliability sets Neva apart from other search engines. The disruption Neva brings to the ad-supported publishing ecosystem poses challenges for publishers, but it also presents opportunities for publishers to rethink their approach and provide more engaging experiences. Privacy is a key aspect of Neva, as it respects user preferences and provides a calm and tracking-free search experience. Ultimately, Neva's goal is to offer a superior product experience by providing answers instead of links, listening to user feedback, and continuously improving their search technology.

Summary & Key Takeaways

• Neva is a search engine and chatbot powered by AI technology.

• The CEO and co-founder, Sridhar Ramaswamy, aims to create a new model of search that focuses on user experience and privacy.

• Neva integrates language models into its search engine to provide quick, authoritative, and timely information.