EN
As developers, it's easy to get caught up in the technical details of building complex apps. However, when creating innovative AI solutions, it's crucial to step back and ask: what should we be building? How can we ensure our AI projects drive real business value and meet user needs?
In this post, we'll move away from our usual technical discussions. We’ll shift our focus from 'building it right' to 'building the right thing,' exploring the strategic considerations and decision-making processes that can help you create AI solutions that truly make an impact.
Hunting for use cases
Potential for AI assistance
When evaluating potential use cases for your AI application, prioritize tasks in your organization that are repetitive, time-consuming, and frequent. These are the tasks that require significant manual effort, are prone to human error, and are ripe for AI to share in the burden. By shifting parts of these tasks to AI, you can free up resources, cut costs, and boost efficiency — and you might even improve performance!
To identify these tasks, ask your target users about the most tedious and repetitive parts of their job. They'll likely be eager to share their pain points, providing you with a great starting point for applying AI. It’s key to remember that you are likely to be assisting these end users rather than completely automating these tasks for them, too.
Quantifiable impact
When exploring AI solutions, it's easy to get caught up in the excitement of cutting-edge technologies. However, without understanding the financial impact, you risk investing time and resources into a project that may not yield the desired returns. To avoid this, you need to quantify the financial impact of a potential AI use case.
There are two main ways AI can have a financial impact: revenue gains and cost reductions. Revenue gains come from using AI to expand into new markets, sectors, or geographies, or to improve performance by reducing error rates. Cost reductions come from using AI to automate tasks, freeing up labor hours and reducing costs.
To quantify these benefits, ask yourself questions like:
- How many hours or employees are freed up by assisting with a task?
- How often is the task performed?
- What is the labor cost per hour?
- What is the size of the untapped market that AI can help you reach?
- How quickly can AI help you reach it?
This exercise may seem daunting, but a mere order-of-magnitude estimate will usually suffice. A few rough-but-realistic assumptions can quickly inform you which use cases are worth pursuing. And then aim for the use cases with a potential value of $10,000,000 per year before the ones worth $10,000 per year!
Human-AI Interaction and Collaboration
A final concern revolves around how (and how often) the end users will collaborate with the AI-based application. An application that replaces a human and/or removes them from the task or decision-making process will be met with distrust and poor adoption. On the other hand, an application that augments the end user’s capabilities tends to be much more welcome.
By combining human creativity with AI's analytical capabilities, you can uncover new insights and create a more intelligent, intuitive, and user-centric app. This approach also leads to more sustainable adoption rates, as users are more likely to trust and adopt a tool when they're involved in the decision-making process and understand the reasoning behind AI-driven recommendations. By embracing human-AI collaboration, you can set your app apart from the competition and unlock the full potential of AI.
An example use case: Knowledge Assistant
With the above features in mind, let’s explore a sample use case for an AI-driven app:
The knowledge bases we rely upon to do many jobs are vast and growing. Whether we are talking about software documentation, user manuals, legal case law, medical studies, or more, it’s difficult for a person to stay on top of changes and to know where the information that they need resides.
AI is a natural assistant for such knowledge-driven tasks. It can explore the knowledge base, flag whenever something has changed or is inconsistent, summarize its findings, and refer the human to the root of the information for confirmation and citation.
This increases the efficiency of research and support teams while also allowing them to explore more knowledge bases than ever before. It can furthermore provide a natural feedback loop to the creators and maintainers of the knowledge base whenever something looks amiss. All while reducing the tedium of repeated, manual searches!
Example Application: SambaNova Financial Knowledge Assistant
We can see just how such an application ticks all three of the boxes we’ve described above:
- Sifting through stacks of documentation, references, databases, and other knowledge sources is tedious, repetitive work. Locating the knowledge isn’t even the goal of the end user: it’s figuring out how to apply that knowledge to the query at hand that’s important.
- Users aren’t cut out of the decision-making process at all. Instead, they just use the AI to gather all potentially-useful information quickly and then skip directly to the interesting part of their role.
- The impact is immediate and easy to measure:
- Perhaps the efficiency of the users is improved by 5, 10, or even 30%, which allows them to take on more cases within the same amount of time
- Perhaps, too, the quality of the users’ research is improved because the AI assistant has considered twice or three times as many data sources as the human alone could have done
- We can take these time savings or quality improvements and multiply them by the team size and the hours spent to arrive at a rough estimate of the value that this assistant creates
Conclusion
The guidance here is not meant to be exhaustive or universally applicable, but we do hope that it helps application developers to think a bit beyond the technology that their apps use and to also consider the people who might benefit from what is being created.
By taking a more user-centric point of view, developers can expect to see stronger adoption and a much larger impact from their hard work!
