During its Inbound customer conference held in Boston today, HubSpot, the marketing and CRM platform based in Boston, unveiled an extensive AI strategy named HubSpot AI. While HubSpot, similar to other enterprise SaaS providers, has been incorporating AI elements over the years, the emergence of ChatGPT this year has prompted them to adopt a more comprehensive approach while retaining some of their earlier AI components.
Dharmesh Shah, the company’s CTO and co-founder, believes that AI, especially generative AI, will fundamentally transform how we interact with software. He stated, “We haven’t witnessed such a significant paradigm shift in software for a long time. Generative AI is poised to revolutionize product development and value creation for customers.” HubSpot has been actively aligning itself with this transformative shift.
Andrew Pitre, EVP of product at HubSpot, explained that the objective of HubSpot AI is to simplify the tasks of marketers and salespeople. The company leverages a diverse range of content, including customer-generated blog posts and stored data within HubSpot, to train its AI models. Additionally, it utilizes this wealth of content and data to comprehend the context of user requests based on existing information.
Pitre elaborated, “So, we gain an understanding of how your business operates, and we’re increasingly working on the ability to grasp your business’s tone and communication style.” Furthermore, HubSpot’s CRM database contains data about companies, enabling users to gain a deeper understanding of their customers, including basic information, purchase history, and customer service interactions.
HubSpot AI comprises four key components. First, there are content assistants, which assist users by generating various types of content within the platform, such as blog posts, images, marketing campaigns, and websites. Second, AI agents are designed to provide automated assistance to customers, handling inquiries and routine customer service tasks. Third, AI Insights offer predictive capabilities, providing insights into customer behavior, such as identifying potential churn risks or high-potential buyers, a feature already integrated into the platform for some time. Lastly, ChatSpot, introduced in March and currently in beta, allows marketing and sales professionals to interact with content stored in HubSpot using natural language queries, similar to ChatGPT.
Pitre emphasized that by limiting the AI models’ access to HubSpot’s data, they can help prevent hallucinations, where the model provides inaccurate information when it doesn’t know the answer. He stressed the imperfection of AI-generated information and advised users to view it as a helpful input rather than absolute truth.
Regarding product availability, AI Insights are already accessible, ChatSpot is in public beta, AI agents will be available next year, and some AI assistants like the content assistant are generally available, while the image creation assistant is in beta testing.
