I’d like to suggest that HubSpot develops its own large language model (LLM) that runs natively inside HubSpot. The model would stay fully contained within each customer’s HubSpot account and only train on the customer’s own data.
This approach would deliver two key benefits:
Privacy and security: All data remains within the customer’s environment, eliminating exposure to external systems.
Relevance and accuracy: The model is fine-tuned exclusively on the customer’s data, ensuring outputs are tailored to their business context.
By embedding an LLM directly in HubSpot, customers could leverage powerful AI while maintaining complete control of their sensitive information.
