OrganAI.se - An app to help you organize.

What problem are we solving?

Scheduling meetings is annoying as f**k.

How many hours have you spent in your life scheduling meetings with others? When we did our research, we learned that it takes an average of 8 hours to schedule a single meeting.

That’s very frustrating and time-consuming as we all have better jobs to do, like working in Figma, preparing for meetings, managing stakeholders... but often, we end up just scheduling meetings.

Back in 2019, in a world without ChatGPT, we decided to train and build our own AI assistant to make scheduling meetings with each other much easier.

Our starting point

We started with a prototype developed by a freelance designer, which, in retrospect, was poorly designed. Nevertheless, this was our starting point.

The initial prototype that we had (in Invision)

How I push for validating the idea

To validate our idea, we turned the initial prototype into a working app and introduced it to potential users. Here’s what we learned:

On the positive side, we confirmed that scheduling meetings is indeed a time-consuming task.

However, no one wanted to use our app due to several issues:
• No one wants to spend time typing and interacting with the AI.
• The feedback from the AI is unclear.
• Our NLP is super bad (obviously... as we didn't have that much training data at that point).

The MVP that we built

Iterating on the feedback

Driven by user feedback, we conducted a workshop to brainstorm how we could make the app more user-friendly.

We developed two new approaches:

  • A button-based interface where users could book meetings with simple taps.
  • A text-based interface where users could type any thing to book meetings.

We built prototypes for these ideas and tested them again.

Text-based vs button-based

Introducing auto-complete, for composing messages

After weighing the pros and cons of each concept, we decided to combine the best elements of both. We retained the typing interface, which allows users to flexibly type in their requests, while simultaneously introducing a smart composer to help users type faster.

This overall experience has helped us increase the System Usability Scale (SUS) rate by 45% and significantly enhanced user satisfaction.

Demo of the auto-complete feature we developed

Final Touches

After integrating the feedback, we polished the UI for consistency and enhanced user experience. The final design was then passed to the development where I led 5 developers to complete the implementation.

Reflection

The reason I am writing this story is, admittedly, because we failed to drive any growth with the app. Despite this, it was a fun journey and a valuable experience to build an app and launch it into the market, where we also managed to raise $500K from one of the startup incubators in Sweden.

If you are interested in learning more about this story, I am happy to share it during our calls.

Read next

Github for Data Scientist