Unlocking Real Value with LLMs in Product Development
Technology
•
Sep 21, 2024
In just a few years, Large Language Models (LLMs) have moved from research labs into the center of the world’s fastest-growing products. Yet for many founders, the question remains: how do you get real business value from LLMs, instead of just another AI demo?
The shift is more urgent than ever. According to a 2024 Gartner survey, over 80% of product leaders say that their customers now expect LLM-powered features—whether that means advanced search, smart recommendations, or instant content generation. The competitive pressure is real: Salesforce’s recent “AI in Product Development” report found that products using LLMs to improve user experience are 50% more likely to reach their growth targets in the first 18 months.
But success doesn’t come from just adding AI and hoping for the best. The most effective teams start by identifying the true bottlenecks and opportunities in their product. For example, Notion and Grammarly have leveraged LLMs to rewrite and summarize content, transforming user workflows in ways that create both delight and loyalty. Stripe built an LLM-powered help assistant that cut customer support ticket times in half, according to their engineering blog.
One lesson stands out from all the research: value comes from integration, not just features. McKinsey’s 2024 State of AI in Product study revealed that companies who embed LLMs deeply into their workflows—such as automating onboarding, enhancing customer insights, or powering dynamic personalization—see the biggest return on investment. These products are not just more efficient; they feel almost magical for users.
However, building value with LLMs isn’t only about technology. Teams need to invest in good data hygiene, clear product thinking, and user-centric design. The risk of bias or “hallucinations” is real—OpenAI’s own usage studies recommend strong human feedback loops and transparent user controls. The most trusted companies are those that not only use AI, but also show customers how and why it’s working.
Perhaps the most important takeaway is that real business value emerges when LLMs are treated as core building blocks, not afterthoughts. As Ben Horowitz wrote in his analysis for Andreessen Horowitz, “The next generation of products won’t bolt AI onto the side—they’ll make AI the backbone.” Whether you’re automating research, building smarter search, or helping users write, the companies that win will be those who use LLMs to reimagine what’s possible, not just automate what already exists.
References:
Related insights
Unlocking Real Value with LLMs in Product Development
Technology
•
Sep 21, 2024
In just a few years, Large Language Models (LLMs) have moved from research labs into the center of the world’s fastest-growing products. Yet for many founders, the question remains: how do you get real business value from LLMs, instead of just another AI demo?
The shift is more urgent than ever. According to a 2024 Gartner survey, over 80% of product leaders say that their customers now expect LLM-powered features—whether that means advanced search, smart recommendations, or instant content generation. The competitive pressure is real: Salesforce’s recent “AI in Product Development” report found that products using LLMs to improve user experience are 50% more likely to reach their growth targets in the first 18 months.
But success doesn’t come from just adding AI and hoping for the best. The most effective teams start by identifying the true bottlenecks and opportunities in their product. For example, Notion and Grammarly have leveraged LLMs to rewrite and summarize content, transforming user workflows in ways that create both delight and loyalty. Stripe built an LLM-powered help assistant that cut customer support ticket times in half, according to their engineering blog.
One lesson stands out from all the research: value comes from integration, not just features. McKinsey’s 2024 State of AI in Product study revealed that companies who embed LLMs deeply into their workflows—such as automating onboarding, enhancing customer insights, or powering dynamic personalization—see the biggest return on investment. These products are not just more efficient; they feel almost magical for users.
However, building value with LLMs isn’t only about technology. Teams need to invest in good data hygiene, clear product thinking, and user-centric design. The risk of bias or “hallucinations” is real—OpenAI’s own usage studies recommend strong human feedback loops and transparent user controls. The most trusted companies are those that not only use AI, but also show customers how and why it’s working.
Perhaps the most important takeaway is that real business value emerges when LLMs are treated as core building blocks, not afterthoughts. As Ben Horowitz wrote in his analysis for Andreessen Horowitz, “The next generation of products won’t bolt AI onto the side—they’ll make AI the backbone.” Whether you’re automating research, building smarter search, or helping users write, the companies that win will be those who use LLMs to reimagine what’s possible, not just automate what already exists.
References: