Description
Summary 1. Overview JMUP is a SaaS web application that lets professionals use online questionnaires generating personalised, AI-produced reports. Users purchase credits; one credit entitles the user to one report. The service is positioned as a decision-support tool: no wording in the interface or in the reports may present a result as an automated decision about a person. This document describes the functional requirements of the target application. It is deliberately technology-agnostic: candidates are invited to propose the stack they are most efficient with. 2. Users and accounts Two account types coexist. The individual account: a user signs up, buys credits and uses them for themselves or for third parties. The multi-user company account: an organisation buys credits in volume, allocates them among several users attached to it, receives consolidated invoicing and has an internal administrator. Not all company features will necessarily be exposed at launch, but the data model must be designed from the outset to host several users under a single entity, so as to avoid rebuilding it later. 3. User journey The primary requirement is simplicity, including for non-technical users. The usage mechanism is as follows: a user spends a credit either by completing the questionnaire themselves, or by sending a usage link to another person who completes it. In all cases, the result always returns to the ordering user, the one who committed the credit: on screen, and as a downloadable, non-editable PDF. Confidentiality of results is a strong principle. By default, no result is stored. The only exception is a short, fixed-duration technical safety window whose sole purpose is to allow a report to be regenerated in case of an incident, after which the data is deleted. As the catalogue will grow, users must be guided towards the questionnaire that fits their need rather than left alone in front of a list. 4. Business model Sales are credit-based, paid by card through Stripe, with compliant invoices issued. Three contractual rules apply: credits must be used within twelve months of purchase, no refunds are offered, and the terms of sale must be accepted before any payment. Discounts and promotional codes applied to a purchase rely on Stripe's native features. Granting free credits without payment, in the form of trial links, is application logic and must be manageable from the admin area. 5. Scope of the assignment (phase 1) The assignment covers the first marketable version, comprising the following. User accounts with secure authentication, including sign-up, login and password recovery. Credit purchase by card through Stripe, with reliable server-side handling of payment confirmations and invoicing. One to two questionnaires available, with AI report generation performed exclusively server-side: the AI service key must never be exposed client-side. Credit consumption either directly or through a usage link, with results returned to the ordering user on screen and as a non-editable PDF, with no storage beyond the safety window. A minimal admin area allowing promotional codes and free trial links to be generated. Acceptance of the terms of sale at checkout. A French-language interface, with a structure ready to host additional languages. An account model ready for the multi-user company logic described above. 6. Planned evolutions (outside scope, for context) Later phases will include regular additions to the questionnaire catalogue, at an envisaged pace of every two to three months, as well as community features, additional languages, a partner area and mobile applications. The design consequence is central: the architecture must be modular, so that adding a questionnaire is a simple operation carrying no risk for the existing system. This capacity for frequent additions is not a comfort but a core product requirement. 7. Technical organisation of the assignment The choice of stack is open: any modern technology is suitable, and candidates are invited to propose the one they master best, with a brief justification. The assignment prioritises back-end skills: application logic, database, security, payment, server-side calls to the AI service. A working prototype of the interface exists as a single-page web application. If the contractor prefers to focus on the back-end, front-end integration can be handled by our technical lead; this point will be agreed at kick-off. The code is hosted in a private GitHub repository owned by the client. Work is carried out on branches and submitted as pull requests, reviewed and approved by our technical lead before merging. Delivery is organised in milestones, each subject to approval. 8. Security and compliance The service targets a European market and must comply with the GDPR: data hosting within the European Union is preferred, results are not stored beyond the safety window, account deletion must be possible, and the system's operation must be sufficiently documented. Keys and secrets (payment, AI service) are kept exclusively server-side, as environment variables, and never appear in the code or client-side. Legal texts (terms of sale,