Respirit.Me

Mental Health ~ Poland ~ 5 employees

Precise AI recommendations on a well-being platform

Challenge:

Users were lost in hundreds of methods and specialists – they didn't know where to start and who to trust

Solution:

Implementation of a diagnostic test based on the OpenAI API, which recommends work methods and specialists on the platform.

5 minutes

This is how long it takes on average for a user to receive a recommendation

Recommended specialists make it easier to make decisions

Greater chance of booking your first session

Our approach

First, we gathered knowledge from Respiritme experts about the most common problems users encounter and the methods that work best in specific situations. This led to the creation of a map: "User type – dominant needs – appropriate methods – specialist type."

Next, we developed a test structure: quantitative questions (scales, symptom frequency, functioning style) combined with qualitative questions (description of the situation, experiences with previous self-improvement). The OpenAI API was used to interpret open-ended responses – the model was able to capture hidden needs, fears, level of readiness for change, and preferred work style.

The results of the responses create a user profile, which is then run through an internal scoring system. Based on this, the test generates recommended job directions and specific specialists.

The whole thing has been designed so that after completing the test, the user has a sense of clarity: they know where to start, who to meet with and what the next step is, instead of continuing to wander around the platform.

Solution

01 Smart test instead of static survey

Based on the user's responses, the test creates a coherent picture of their situation and priorities. Instead of a series of random questions, it guides them through a logical path - from symptoms and habits, through previous attempts at self-improvement, to their preferred support style - and ultimately translates this into specific recommendations for methods and specialists.

02 Response mapping

Test responses - both scales and open-ended responses - are translated into a clear profile: dominant areas of stress, stress response style, energy level, and readiness for self-improvement. This profile serves as the basis for further recommendations.

03 Understandable recommendations

Based on the profile, the test generates a list of methods and specialists, but not in a dry table format. Each recommendation is accompanied by a short, understandable justification: it explains why a given method can help and what to expect from the first sessions. This reduces the fear of new approaches and lowers the barrier to entry.

04 App integration

The test result becomes the starting point for the entire Respiritme journey. From the summary, users can: book a session, save the results to their profile, return to them later, or compare them with subsequent tests over time.

Results

Conversion increase

People who completed the test were much more likely to schedule a first appointment because they already had data and a personalized recommendation, rather than a general list of services.

A greater sense of agency

Test participants feel like they are taking part in a diagnostic process, rather than purchasing a random service.

Fewer support inquiries

Clear recommendations after the test meant that users were less likely to write asking "who should I contact?"

Better use of the product

Thanks to intelligent recommendations, users are directed not only to the most obvious categories, but also to complementary methods that actually support their situation.

Greater user confidence

After the test, users receive a clear suggestion for the first step – a specific method and specialist – instead of long-distance browsing through a list of options without making a decision.

Better fit

The test reduces the number of “unsuccessful” first sessions in which the user’s expectations and the specialist’s work style simply do not match.