Building Product with AI: From Insight to Production

In today's organizations, ideas are not lacking; the real challenge is turning them into a real business impact. With the advent of Artificial Intelligence, the speed of prototyping and development has multiplied, but this brings a risk: moving too fast and losing focus on whether what we build truly adds value.

In our recent Runroom LAB: Building Product with AI, our colleagues from Runroom Jorge Agúndez, Product Owner, Laura Polls, Head of Experience Research and Director of Runroom Academy and Jose Saura, Head of AI, shared how we are transforming research into AI-accelerated product decisions. Through the real case of Grupo Indukern, they broke down how to go from an insight to a product in production in record time, combining strategy, Lean methodology, and agentic flows.

Below, we share the key learnings from this session:

The Indukern case: Impact in 5 weeks

The challenge was clear: transform an inefficient internal service (a pharmacy counter with discounts for over 1000 employees) into a useful, scalable digital product that reduced operational load.

To achieve this, we avoided large waterfall projects. We sized the effort in just five weeks: one week of express Discovery, one of setup (Sprint 0), and three of Delivery. The methodological key was not to treat these phases as isolated compartments, but to integrate discovery and delivery into a single continuous flow. The learnings directly impacted what the team was building.

Discover how we moved from a manual process to a digital employee counter in just 5 weeks in the success case.

The "Meat Brain" and AI as a Cognitive Companion

To meet such aggressive deadlines without working "without a net," we heavily relied on AI (using tools like NotebookLM and Claude). However, the role of technology was very well defined compared to that of the human team.

As Jorge highlighted during the LAB: "Use AI to accelerate the process but not to replace it in decision-making."

AI took care of the heavy lifting: structuring qualitative data, analyzing interview transcripts, generating mind maps, and identifying frictions. But the "meat brain" (our research and product team) was responsible for empathy, context interpretation, and strategic alignment with the business.

Focus on Outcomes: Behavioral Changes over Solutions

A common mistake when creating products is to immediately think of the final solution (tangible functionalities). In this project, the focus was on Outcomes, that is, the behavioral changes we wanted to generate in employees.

During the impact workshop, the business wanted employees to memorize the company's vademecum (catalog). However, human research (and conversations generated through a quick artifact created for validation) showed that users only wanted an experience similar to e-commerce: easy to buy, at a good price, and receive it quickly. Focusing on outcomes allowed us to discard unnecessary functionalities (like a purely visual catalog) and invest the little time available in what truly moved the business needle.

Until now, the bottleneck was development, and now the bottleneck is the business.

Jorge Agúndez,
Product Owner, Runroom

From GPTs to "Skills" and Agents: The Technical Evolution

AI tools evolve at a dizzying speed; in fact, what we did manually with prompts a few months ago is now addressed automatically.

During the LAB, Saura showed us the qualitative leap: moving from using generic chats to building Skills. A Skill is nothing more than a markdown file that packages instructions, resources (context files), and deterministic scripts to standardize a specific skill.

For example, instead of copying and pasting interviews each time, a Skill for research synthesis can automate analysis, cross insights and systematically draft user stories. Going a step further, these Skills feed subagents that can operate in parallel: one agent analyzes the competition, another drafts, and another programs code tests.

The New Paradigm: The Bottleneck Has Changed

The adoption of these technologies leads us to a final reflection that emerged during the LAB debate and changes everything in digital product creation:

"Until now, the bottleneck was development, and now [...] the bottleneck is product [or business]."

Thanks to tools like skills and agents, code writing and data processing have accelerated dramatically. Technology is no longer the limit. Now, the real challenge lies in product validation: understanding what we should build, discovering the right problem, and ensuring we are delivering real value to the user before we start developing.

At Runroom, we continue to explore this fascinating intersection between human strategy and artificial acceleration. Because at the end of the day, AI can build the wheel in seconds, but we are the ones who must decide where we want to roll.

Runroom Academy

Service Design & Prototyping Accelerated by Generative AI

If this tension sounds familiar (increasing speed, slow validation), the answer is not “more tools,” but a method to turn AI into a strategic asset: reduce uncertainty, prioritize with impact, and prototype to learn first.

The in-person bootcamp “Service Design & Prototyping Accelerated by Generative AI” by Runroom Academy is designed exactly for that: integrate AI into decisions, shorten validation cycles, and lower innovation risk.

Next edition: May 14 and 15 in Barcelona, with an informative webinar on March 25, and early bird until April 3.

More information and links to register for the webinar and training here

You might also be interested in:

• Training in Service Design & Prototyping Accelerated by Generative AI | Runroom Academy