What we delivered
We delivered a complete product for a startup founded by a group of psychologists. The engagement involved close work with both the startup team and the accelerator supporting it. The resulting application was released in Polish and English, and the startup began testing it with users.
The product helped psychologists manage notes about patients and revisit the context of earlier sessions. It was intended for professionals working in private practices as well as public care. Voice input reduced the friction of recording observations, while a conversational workflow supported therapist case reflection without attempting to make clinical decisions.
The continuity problem
Psychologists accumulate observations across many sessions and patients. Notes can preserve individual facts, but they do not automatically help a professional recall the full context of a case or notice which areas deserve another look. The product addressed that memory and organization problem.
At the time, combining longitudinal patient notes with guided reflection was an unusual product direction. The aim was not to add artificial reasoning to clinical work. Instead, the case reflection assistant organized information already provided by the psychologist and used questions to support a more deliberate review.
That boundary shaped the product. It was a conversational tool for therapists, not an automated supervisor. It did not replace diagnosis, treatment planning, risk assessment, or the psychologist’s clinical responsibility. Its role was closer to coaching and prompting: helping the professional examine the available context and consider whether bias or an overlooked perspective might affect their view.
Product and mobile experience
We built the service as a web application with a Progressive Web App experience for mobile devices. This architecture fit the startup’s budget while still supporting the workflow on a phone. The team did not need to fund separate native applications before testing whether the product solved the intended problem.
The mobile flow mattered because notes and reflections needed to be accessible around the psychologist’s working routine. Voice input provided another way to capture material when typing was inconvenient. The application worked in both Polish and English, so language support was part of the delivered product rather than a later translation exercise.
Patient notes were encrypted. We also tested local models as an alternative processing direction. These decisions responded to therapy data privacy and the sensitivity of patient-related material. The available evidence does not identify the encryption algorithm, prove a specific regulatory certification, or establish that every processing path remained on a local device. Those claims are therefore outside this case study.
Collaboration and iteration
The engagement involved two stakeholder groups: the psychologist-founded startup and the accelerator it was participating in. We worked with both during product development, while keeping responsibilities clear. The startup’s psychologists supplied the domain knowledge and shaped the professional boundaries of the workflow. The accelerator formed part of the wider stakeholder environment around the project, without replacing the founders as the source of product and clinical decisions.
The work covered product design and implementation, not only a technical demonstration. We had to balance note management, continuity between interactions, voice capture, multilingual use, privacy controls, and a mobile experience within the available budget. Choosing a PWA was one of those product decisions: it concentrated investment in one responsive application while preserving a practical installation-like experience on mobile.
Clinical review remained separate from software behavior. A useful prompt could invite the psychologist to reconsider an assumption, but the product could not determine whether that assumption was clinically valid. The professional remained responsible for interpreting the notes and deciding what to do next.
Supported outcome and limits
The supported outcome is a full bilingual product released by the startup for user testing. It included encrypted note management, voice input, guided case reflection, a responsive web application, and a mobile PWA. The product supported both private-care and public-care psychologists.
This account does not claim improved patient outcomes, unbiased clinical decisions, diagnostic accuracy, time savings, regulatory compliance, or adoption beyond the reported testing stage. No quantitative user-test results or later operating status were supplied. It also does not describe the application as clinical supervision.
For a related product, the practical starting point is to define the permitted task, professional boundary, data flow, and evaluation method before selecting a model. Read more about our approach to conversational AI design.