Leverage LLMs for Fast Value
Quick wins = long term success
Many product leaders are looking to AI for “magic” to solve business problems with a silver bullet. They’re looking for the next big opportunity to introduce transformational change and leave their mark on the organization. The hope? To find that breakthrough, that disruptive innovation, which will redefine their business trajectory.
But usually, the best way to create tangible change is start with incremental quick wins. It may seem obvious, but all too often, you see sweeping, 3-to-5 year visions created by organizations result in disappointment due to the inability to show traction early and often in the journey.
The current AI revolution is no different, and the most successful approach right now is the one in which you can create “fast value”.
Why fast value?
The main goal is to build internal trust. It’s likely your executives and stakeholders have been burned by AI before. Maybe a few years ago, prior to the explosion of LLM models like GPT and LLama, they’ve invested in traditional predictive models maybe that have showed business value when implemented against 2–3 use cases, but as customers evolve, so do those uses cases. And when they try to expand those 2–3 uses cases to 10 — the technology is slow to scale, and those leaders find them selves burndened with huge implementation bills from their product and IT departments. It’s important to build trust, show that your use cases work and demonstrate its flexibility — which is after all, the power of LLMs.
Secondly, implementing AI is not a zero-sum game. Very successful long term business outcomes often come with short term challenges as the business shifts to tweak and adjust their strategy. As you start to build AI applications around your own business’ data and use cases, you’ll uncover alot of things. You’ll find what your true points of failure and learning are, and you’ll have a much better understanding of constraints and possibilities.
Put these together, can you put yourself in a position where you build organizational momentum behind your AI efforts.
Use Cases and Examples
Consider these use cases and examples for how you can create fast value with AI.
Addressing UX Research Friction
Almost every design organization grapples with the challenges of UX research. Traditional methodologies are resource-intensive, time-consuming, and often struggle to scale at the pace of your business. AI techniques can help scale your existing research to extract value faster, creating richer, more diverse inputs for the design processs. Some techniques include:
- Synthetic data — leverage generative AI to mimic user behavior data in order to augment your existing research — giving you larger data sets that would otherwise cost more more resources. One study shows that leveraging LLMs creates a savings of 20 times more than using other surveys and data augmentation techniques like crowdsourcing
- Topic Clustering and Sentiment Analysis- leverage NLP techniques to rapidly analyze large amounts of qualitative data for emerging trends, patterns, or even sentiment analysis in user feedback, reviews, or comments in nearly real-time.
- Summarization tasks — Ensure more consistency in how research is interpreted and summarized, reducing the time it takes for human summarizations
Like all research-oriented techniques, using AI doesn’t mean you can take the outputs all at face value, and you still need to experiment with different methodologies on your organization’s data, and interpret the outcomes and correct for biases. A recent study at the University of Chicago showed the promise of employing LLMs to augment research- with a thoughtful and nuanced approach, they can be used responsibly while minimizing the risks.
The tangible win? You’ve just reduced research times and successfully brought new ideas into prototyping in weeks instead of months — building a new muscle for the organization to prepare it for an AI-enabled workflow.
Enhancing User Search Experience
Traditional keyword-based search functions often provide a linear, limited user experience. However, semantic search — especially when powered by AI — enlarges the scope of search exponentially, not only creating a much better experience for its users and creating brand love, but also creating a way for brands to understand user intent like never before.
Platforms like Algolia provide third-party solution that plugs-in to your data and a way for companies more rapidly make an impact with semantic search capabilities.
To get started — first think about your goals and prioritize the top areas that drive the most revenue for you. Structure and tag data in the prioritized areas, and integrate semantic search. Ensure you invest the right resources in tweaking the results and the search experience to drive value for those users. Once you’ve shown some ROI in one area, you can start expanding and integrating semantic search across your wider user experience.
The Silent ROI: Breaking Operational Silos
One of the less discussed, but profoundly impactful applications of AI is in collaboration between different disciplines. Large companies have built very entrenched silos that are now being challenged by the potential of AI. Leaders in the space are investing in multi-disciplinary teams and a culture of cross-department learning and experimentation.
By investing in these collaborations, you can break operational barriers. Try this exercise: Pair a small team of data scientists and creative technologists with a business subject matter expert around a specific product pain point, such as low completion rates for a user task. They may co-collaborate on understand how to best use AI applications to solve the pain point. They may explore how to fine-tune an LLM to offer better customer support via chat, or how an embedding model can leverage their internal company documents to provide more nuanced, contextual information at the right time. Then bring in a UX specialist to generate new ways for this customer to interact with your product.
This creates an environment where your business leads better understand the technology and the capabilities, and data experts better understand the customer and business realities, and it creates enough of a opportunity definition for designers to create a solution that adds value.
In effect — you not only generate new ideas, but you’re also training your organization how to solve problems within the context of new AI technology. Wether or not your ideas can be implemented in 1 month or 12, the immediate value you create proving out a collaboration framework for AI transformation.
Conclusion
The potential of LLMs is vast, but to stay ahead in this space, consider striving for fast value over large transformations. In this nascent stage, ideas are commoditized, but being able to demonstrate momentum will help you win in the long run. Remember these key principles:
- One of your most important and early milestones in long-term AI transformation is building internal trust
- Build momentum by understanding and prioritizing a pain point that can be solved by AI, identifying the right solution, and crushing it on repeat
- Fostering a culture of collaboration between teams is the silent ROI of AI innovation
