VentureBeat
Follow
HubSpot’s Dharmesh Shah on AI mastery: Why prompts, context, and experimentation matter most
HubSpot's INBOUND conference in San Francisco focused on marketing, sales, CX, and AI innovation. HubSpot CTO Dharmesh Shah discussed AI's dual nature as an exponential opportunity and existential threat. He emphasized collaborating with AI rather than competing against it, viewing it as a tool with immense predictive capabilities. Shah highlighted that AI's learning curve is linear compared to its exponential capability growth. He likened generative AI to having thousands of PhDs in your pocket, enabling creative and expansive tasks. However, he noted AI limitations like reliance on training data and occasional hallucinations. Shah offered a pro tip to try AI for every computer task, viewing failures as temporary setbacks. He stressed the importance of prompt engineering, including prompt quality, model choice, and context. Beyond prompts, context engineering with custom instructions significantly improves AI results. Shah declared this the decade of AI agents, not just the year, with millions already using platforms like Agent.ai. He proposed the TEAM strategy for bringing AI to teams: triage, experiment, automate, and measure. Ultimately, humans win with emotional quotient, merging lived experience with AI for augmented intelligence. AI is intended to handle repetitive tasks, allowing humans to focus on more remarkable and joyful work.
