From concept to code: this is how you can build your first MVP with AI

The early stages of a startup are usually marked by the need to validate hypotheses and materialize ideas without yet having a technical team. In this context, having an MVP becomes an essential step to verify whether there is real interest, collect feedback and move towards a more complete product. Generative AI and code agents are transforming this initial phase, allowing functional prototypes to be created quickly, economically and accessibly.

With this purpose, the ONE Platform held the webinar “How to create the first MVP of your startup with artificial intelligence”, a session aimed at entrepreneurial people who want to start building their project without having extensive experience in programming, beyond basic knowledge to install the necessary tools. Marco Gonzalo, co‑founder of 4Geeks Academy Spain, shared practical insights on code agents, good practices for working with AI and a real case of building an e‑commerce MVP.

Code agents: what they are and how they can help you create an MVP 

The first block was dedicated to understanding what a code agent is, how it works and why it has become a strategic tool in the process of creating a digital product. A code agent is a system powered by AI capable of understanding development goals, planning steps, writing and modifying code, executing actions and evaluating results to iterate autonomously.

During this part it was explained that not all agents are the same: they vary in their level of autonomy, in their ability to execute complete tasks and in the environment in which they operate. Marco detailed the differences between the main agents available, the environments where they can be used —IDEs (integrated development environments), CLIs (command‑line interfaces) or cloud services— and their advantages and limitations depending on the type of project.

This analysis allowed attendees to understand when it makes sense to use a code agent and what realistic expectations they should have in their first tests.

How to work correctly with a code agent

One of the central messages of the webinar was that the value does not lie solely in the tool, but in how you work with it. To this end, three essential concepts were explored in depth: supervision, context and rules.

First, emphasis was placed on the human‑in‑the‑loop approach. The entrepreneur must guide, validate and correct each iteration. AI can accelerate technical execution, but it does not replace decision‑making or the strategic definition of the product.

Secondly, the concept of context was explained in detail and how it directly influences the quality of the result. Context includes the product description, technical requirements, the current state of the project, expected deliverables and existing constraints. It was also clarified what should not be considered context: sharing unstructured ideas or informal thoughts does not help the agent produce better results. Clarity and specificity are key.

Lastly, the role of working rules was addressed. Unlike context, which describes the project, rules define how the AI should behave: what steps to follow, which standards to apply or which constraints to respect. Writing clear and coherent rules allows for greater consistency in development and facilitates future iterations.

Practical case: creating an e‑commerce MVP with AI

The central part of the webinar was a live demonstration where Marco built a simple e‑commerce MVP applying what had been learned. The exercise included defining the product statement, an initial landing , a basic catalog, a cart and a simulated order flow.

The exercise began with defining the implementation plan and structuring the project. From there, clear rules were established and the AI was given the necessary context to begin development. Throughout the process, components were validated, errors corrected and successive adjustments made until a functional result was achieved.

This exercise made it possible to see how a project that would normally require hours of development can be built in an agile and understandable way even without advanced technical knowledge. 

A boost to create your first product

The session concluded with a clear message: AI does not replace entrepreneurial judgment, it amplifies the ability to move forward. Thanks to code agents, entrepreneurs can validate faster, reduce costs and materialize a first product that allows them to learn and evolve.

The closing invited attendees to continue exploring training content on the ONE Platform and to take advantage of upcoming sessions to continue developing technological skills applied to entrepreneurship.

Do you want to dive into all the details? You can watch the full webinar video and follow the recommendations step by step. 

If you don’t want to miss the upcoming sessions, check the webinar schedule of the ONE Platform.