The future of enterprise search and AI-powered work productivity with Glean’s Arvind Jain | E1916

Glean’s Arvind Jain joins Jason (00:00:00)

• Arvind Jain from Glean joins Jason to discuss the future of enterprise search and AI-powered work productivity.
• Jain believes that the majority of AI work in the future will be based on open-source models.
• Open-source models are hard to beat, and many startups are using them.
• Glean is an AI-powered search engine and assistant that helps people get more work done.
• It allows users to ask any questions they have and Glean will use all of the company's knowledge, data, and information to answer those questions.
• Glean is not a general AI like ChatGPT, it is specifically designed to search and answer questions within an enterprise's data.

Glean’s beachhead market and primary focus (00:02:35)

• Glean is typically deployed company-wide and sold to CIOs.
• The top users of Glean are engineers, support personnel, and sales personnel.
• Glean is useful to every employee in a business and is sold through the CIO.
• Glean sits on top of existing systems like Intercom, HubSpot, and Zendesk, acting as a connective tissue that connects knowledge across different systems.
• Glean helps users find the right answers to resolve cases, even if the answers are in external sources like Slack conversations or internal Jira issues.
• Glean helps support agents quickly find information and resolve customer requests by tapping into internal discussions and conversations.

Glean’s approach to data confidentiality (00:05:13)

• Glean utilizes various language models, including PaLM 2, Gopher, and GPT-4, allowing customers to select their preferred model for summarization and synthesis tasks.
• Glean collaborates with Azure to leverage GPT-4 while ensuring data privacy through a contract that safeguards customers' data.
• Access to information within Glean is restricted based on individual user permissions within the organization.
• Many enterprises trust major cloud vendors like Google, Microsoft, and AWS for secure data storage and rely on their robust security measures.
• Glean supports hosting open-domain language models as the core language model in its enterprise search platform.
• To prevent data exposure, CIOs should ensure employees avoid using public consumer-based products like ChatGPT, which may inadvertently share sensitive information with the provider.
• Glean offers a secure environment for enterprise search, ensuring that data sent to Azure or Google complies with security agreements and is not misused.

OpenPhone - Get 20% off your first six months (00:11:57)

• OpenPhone simplifies business communications with an easy-to-use app.
• It can be used on phones or desktops, making it convenient for teams.
• Shared phone numbers can be created for multiple employees to handle calls and texts.
• OpenPhone is the number one rated business phone on G2 for customer satisfaction.
• Twist listeners can get an extra 20% off any plan for the first 6 months at openphone.com/Twist.
• OpenPhone can port existing numbers from other services at no extra cost.

Competition against native tools and building language models in SaaS companies (00:13:13)

• Glean builds smaller language models to understand company knowledge.
• SaaS companies like Salesforce, Atlassian, and others are not building their own language models but are using APIs to access large language model providers like GPT-3 or Jini.
• In the future, every application is expected to have some AI smarts and may even move to a chat interface.
• Glean solves a different problem by tying together knowledge from different systems in one place, making it easier for users to find information and ask questions.

Glean’s approach to permissions (00:17:00)

• Glean's AI-powered search tool understands permissions and retrieves documents that users have access to.
• Glean indexes content from various systems and continuously updates its index in real time, providing fast search results without the need to send messages to multiple systems.
• Glean does not store any Enterprise data in the language model but uses it for understanding and generating responses.
• Search functionality is often overlooked by enterprise software companies, who focus on competing features.
• To solve the search problem, Glean considers the enterprise context and uses signals from various sources to determine the relevance and importance of information.
• Enterprise knowledge can be viewed as a graph with interconnections between different pieces of knowledge and people.
• Glean leverages enterprise-wide data to understand user behavior and preferences, delivering more relevant information to users.

Lemon.io - Get 15% off your first 4 weeks of developer time (00:24:20)

• Lemon.io offers on-demand developers with at least three years of experience and strong portfolios.
• Only 1% of candidates who apply to Lemon.io are accepted.
• Lemon.io handles the integration of developers into teams.
• Twist listeners get 15% off their first four weeks with Lemon.io.

Compliance and the "CEO God Mode" feature (00:25:23)

• Glean has a compliance mode that is highly restricted and available only to governance and legal teams.
• The compliance mode allows users to search for specific information across all documents, such as insider trading or project acquisition-related keywords.
• Glean is not a system of record, so users do not need to explicitly delete data in Glean when it is deleted in the original system.
• AI can help businesses address governance gaps by automatically finding sensitive information that may have been missed due to incorrect permission settings.
• AI-powered tools like Glean can help businesses improve productivity by automating tasks such as searching for information and answering questions.
• AI can also help businesses address governance gaps by automatically finding sensitive information that may have been missed due to incorrect permission settings.
• AI can help businesses make better decisions by providing insights and recommendations based on data analysis.
• AI can help businesses improve customer service by providing personalized and efficient support.

Potential of productivity software, and people analytics in tech businesses (00:29:28)

• Glean's AI-powered search platform analyzes vast amounts of enterprise data to identify the most productive employees and their work patterns.
• The platform provides insights into individual productivity, including working hours, active resource usage, and potential idle time.
• Glean empowers individuals to gain insights into their own work patterns and identify areas for improvement.
• Managers can use Glean to get summaries of their team's work, but they can only access information that they have permission to see.
• The future of enterprise search involves understanding user intent and providing relevant resources, with conversational interfaces playing a significant role.

Impact of AI on business models (00:40:04)

• AI's ability to directly answer user questions may disrupt the cost-per-click (CPC) business model of search engines.
• As AI provides direct answers, users may be less likely to click on ads, reducing the effectiveness of the CPC model.
• Search engines may need to develop new monetization strategies, such as charging for impressions, clicks, or conversions, or by facilitating transactions and taking a cut of the revenue.
• Google, as the dominant search engine, is well-positioned to adapt to these changes and continue generating revenue from user searches.

Arvind’s thoughts on Google Gemini (00:42:45)

• Arvind believes that Google is in a strong position regarding AI models.
• Despite any issues, Google's AI models, like Gemini, work well.
• Google has the best AI talent, the most extensive data centers, and the necessary machines.
• Arvind thinks the concerns about Google's decline are exaggerated.
• Google's cautious approach to releasing AI models is due to its core mission of providing accurate information.
• Apple's limited release of AI features is likely due to their focus on fit and finish.

Open source vs closed models in AI development and the progress of OpenAI (00:45:17)

• Majority of AI work in the future will be based on open-source models (80%).
• Open-source models have a strong momentum and are hard to beat in terms of technology advancements.
• OpenAI's lead in text-based models is closing every quarter.
• Smaller, faster, and cheaper models are becoming more viable alternatives to costly models.
• It is becoming increasingly difficult to decide which model is the right choice for a specific task.

AGI and the limitations and potential of AI as an assistant (00:47:49)

• Artificial general intelligence (AGI) is when an AI agent completely replaces a human role in an enterprise.
• Glean is an enterprise search and AI-powered work productivity solution that aims to be an assistant for everyone at work.
• Glean's pricing model is per seat, with a focus on midsize to large enterprises.
• Glean's setup is turnkey and can be up and running within a day in a 2,000-person enterprise.
• Glean's domain name, gle.com, is valued at around half a million dollars.