Inspiration

I'm a busy father of 2 toddlers. I have a small business, and some important side projects. I moved to Europe, and I need to learn the local language. I need to stay healthy, while also finding good book, music and movies in the little time I have to unwind.

Some weeks it feels impossible to keep up. Yet, well made AI agents can take care of around 30 to 40% of the tasks I need to do, while it can organize and curate another 10% or more.

This is why I created Prod. A mobile first, AI execution assistant, that can actually do the work.

When I posted a single comment on Reddit about what I'm working on, over 1500 people visited the landing page, and leading to 50+ waitlisted users.

The feedback I got made it clear - there are 10s of millions of people in the same position as me that can benefit from this app.

What it does

When you first open Prod, you give it a goal. Prod then breaks that goal down into project and components. Then, for every component, it creates tasks, and does as much of the work as possible.

However, Prod uses Qloo's insights and taste platform to personalize the goal as much as possible. Studies shows that when you enjoy something, you stick to it longer. Trying to get healthy? Prod can help you find diets and foods you can stick to, especially when shopping for it. Starting a business? Prod can help you do better market research to better understand your customer's affinity's and motivations.

Some key tools Prod has:

• Research - Prod can scour hundreds of webpages for data in the background • Computer Use (Beta) Prod can allow you to scrape information like your recent invoices from a website behind authentication • Coding - Prod can actually write software for you, whether you need to launch a landing page, create a script, or build an app MVP

How we built it

This app is directed to consumers, and therefore, it had to be mobile first. Using react native, we created an ios and android version of the app, and in the future, we will create a web applications as well. Some of the technologies we used include:

• Qloo APIs when the goal has key taste and cultural enrichment aspects • Google's Genkit platform for building the underlying AI agents • Firebase, for backend database and authentication • LLMs - Open AI (o3 + o3-mini) + Google Gemini (gemini-flash-2.0 + gemini-2.5-pro)

Challenges we ran into

The hardest part, which we will continue to work on, is ensuring that the LLM can generate high quality recommendations. While Qloo is great at especially coming at with cross domain discoveries, because we are dealing with consumers with short attention spans who want to enter as little information as possible, designing the UX to require little from the user, but deliver a lot was a challenge. We solved this by giving our AI agent a "clarification" tool, where it can ask follow up questions to better help it determine the best plan and recommendations to make.

Accomplishments that we're proud of

I am proud that in just 1 month, we have a working beta product. We also have over 50+ waitlisted users which are completely random people I have never met before, many taking the time to explain why this solves their pain.

What we learned

Building AI and LLM applications is hard. Although AI might seem to give magical powers, it is a beast that might be tamed. We spent a lot of late nights figuring out the best prompts to get the models to give accurate answers.

What's next for Prod

Prod is now my full time Job. It is genuinely solving a pain, people have shown great willingness to pay, and the market size is large. For the month of August, we will continue to refine our first version based on early user feedback, and launch general availability to the world and iterate from there.

Built With

Share this project:

Updates