Categorías
Asesoría para fundadores

¿Cómo puedes saber si tu idea de software de IA merece una inversión?

Con el creciente entusiasmo de la IA, los fundadores e inversores necesitan validar los conceptos para obtener un valor real para el usuario y evaluar con detenimiento el costo y la seguridad de los LLM comerciales como ChatGPT.
Kelly Gebo
Kelly GeboSenior Associate of Business Development, Thoughtbot
September 9, 2024
How do you know if your AI software idea is investment-worthy?

With all the excitement around AI, there are opportunities as well as pitfalls. Founders and investors need to validate AI concepts to differentiate viable innovations with user value from unrealistic or unhelpful concepts. Additionally, they must be thoughtful about commercial LLM integrations such as ChatGPT that can end up being too expensive or not meeting their security needs.

At thoughtbot, the early discovery and prototyping process is where we’re seeing some of the most valuable work being done to set the stage for successful, lasting AI solutions. As a product agency, we focus on helping clients build the right thing, the right way, with a human-centered design approach. With the AI advancements the last few years, there is no shortage of ideas. We see well-meaning entrepreneurs waving a technical concept around in search of a problem it can solve. Our specialty lies in guiding teams through the process of aligning on the most impactful and feasible solution to real problems for their business and end user, and then rapidly testing that idea.

How do you prove that your AI idea is worth investing in?

To give investors confidence in your vision, and the impact of your AI solution, you need proof. The most compelling proof is done through research and validation. With a mature product, you will track things like engagement metrics to understand your user's experience and product value. But pre-product, it is essential to fundamentally understand your users' core needs and motivations to make sure your product is valuable to them.

Innovations like AI can also bring out complex emotions in users such as fear of the unknown. It is just as important to understand user sentiment as it is technical feasibility. To do this, approach research with empathy, and make sure you understand the emotions of the user at each step.

Not doing validation leads to failing to learn valuable lessons before you and your investors have invested substantial time and money. Validation up front demonstrates to your stakeholders that you are strategic about what you’re building and that you are prioritizing the most simple but effective version you can launch with. This will ensure the best ROI for all parties - including your users!

The best way to validate an AI product concept

Design sprints are the best way to answer difficult questions and find alignment and direction. A design sprint is a highly collaborative, time-constrained process that uses exercises to dig into ideas and assumptions, and quickly prototype and test possible solutions. Over the years thoughtbot has developed a few design sprint iterations to fit different needs and our latest version is focused on AI exploration. The goal of our AI discovery sprint is to uncover if and how AI could be a useful technology for your internal processes or external offering, and then determine the technical feasibility of bringing it to life. 

Like any design sprint, it incorporates rapidly ideating with stakeholders, diverging to explore those concepts, converging on the best options, and testing them with your target users.

When determining the technical feasibility of an AI solution, the initial question to answer is: can this be achieved 1) without AI 2) with existing open source or commercial LLMs or 3) is this a concept that requires custom Machine Learning engineering? Spend time thinking through open source providers, third party solutions, and product launch must-haves, and ensuring you have insight into whether the architecture can scale down the line or if you will hit a limit at a certain point.

In a series of technical feasibility spikes, the team dives into API documentation, experiments with API connectors, and builds prototypes to learn quickly and de-risk the idea, and that can be demoed to stakeholders and tested with real users.

Bringing it all together

There are a few reasons discovery sprints are our go-to way to get started. Not only do they align stakeholders and the implementation team, but they give focus on who to build for and what is built. Completing quality research and testing out various solutions mitigates substantial risk during the formal implementation. As a founder, this helps you understand the runway and investment needed to reach a formal launch and future growth milestones. At the end of the AI discovery sprint you should have: 

  • gained insights from user research

  • validated initial concepts of an idea

  • a Minimal Viable Product roadmap

  • clarified the business impact

These elements are the pillars that can give you and your early-stage investors confidence in your idea and a reliable plan for the future.

Acerca del Autor

Kelly Gebo

Kelly Gebo

Senior Associate of Business Development, ThoughtbotKelly leads Business Development efforts and Client Success efforts at thoughtbot. She loves building relationships and learning about a client's goals and challenges to best support their project.

Suscríbete a The Weekly Index para obtener contenido exclusivo