A web service where employees can adjust career parameters and immediately see the forecast on charts, while HR and administrators manage formulas, dictionaries, and scenario templates through Directus — no engineering involvement. An AI assistant explains the result in plain language, highlights eligibility blockers, and proposes the next steps.
An internal web portal for employees of a large biomedical center. The user signs in, enters starting data, selects a planning horizon, and adjusts career parameters: grade, project role, workload, performance indicators, training, certifications, and threshold requirements. The system recalculates the forecast on the spot and surfaces it as summary figures and trend charts. An AI layer runs on top of the calculated result — it explains what influenced the forecast most, why one scenario performs better than another, and which concrete actions are needed to move to the next level.
The client is a large biomedical research center in the Netherlands focused on translational medicine — from laboratory research through to implementation in clinical protocols. One of its key directions is genetic disease diagnostics, where the path to diagnosis is long and data interpretation is complex. The center works with rare inherited neuromuscular and metabolic disorders, and with selected groups of hereditary immunodeficiencies, combining genetic testing, biomarkers, and clinical observations to shorten time to diagnosis and improve therapy selection or trial participation.
The organization is built around multidisciplinary teams: laboratories, bioinformatics, clinical coordinators, quality and compliance, operations, and HR. Because of the domain, the career model can't be reduced to tenure alone — promotions and permissions are tied to GxP requirements, ethics, training, project participation, documentation quality, audit outcomes, and protocol responsibilities. The center wanted a way to make career-path rules transparent and reproducible for employees, and at the same time manageable and safe for HR and operational leaders to update without constant engineering involvement.
80+ formulas
The platform had to consolidate more than 80 formulas with dependencies, exceptions, and conditions, define the calculation order clearly, and make sure that changing one parameter wouldn't produce unpredictable shifts in the forecast.
A clear user interface
Employees come in with very different levels of digital literacy. The scenario-planning flow had to stay direct and avoid the optional choices that quietly create errors.
Visualization
Charts had to actually help people read trends and compare scenarios — not just dress up the page. Correct period slices, sensible labels, consistent behavior across different planning horizons.
Configuration without developers
The client needed a back office where coefficients, rules, options, guidance texts, and scenario templates could be updated without code changes and without putting the calculation logic at risk.
Reliability and reproducibility
The same input set had to produce the same result, every time. That meant unified rounding, consistent units of measurement, and predictable behavior on unusual or incomplete data.
AI explanations without hallucinations
AI had to explain results and suggest next steps — but it couldn't do calculations instead of the simulator, and it couldn't invent the center's policies. The work needed restrictions, a controlled terminology layer, and answer-quality rules.
We built the service as a managed simulator with centralized rules, transparent configuration storage, interactive charts, and an AI interpretation layer. The formulas live inside a controlled calculation module and return a structured result. The web interface presents summaries and charts. AI operates strictly on top of the already calculated data — explaining what happened, where the blockers are, and what to do next, using the rules and dictionaries stored in Directus.
Web client for career scenario modeling
We implemented authentication, starting-parameter input, planning-horizon selection, variable switching, and side-by-side scenario comparison. The interface updates the result the moment a parameter changes.
Dedicated calculation module with an explicit computation order
We organized the 80+ formulas into a pipeline with a clear sequence — input normalization, base values, derived metrics, final outputs. Validation and human-readable errors went in alongside, so users see the actual reason something doesn't work, immediately.
Forecast visualization with Chart.js
We designed charts for the dynamics of key indicators with support for different planning periods. The backend returns arrays of points and period-based slices, so the charts stay stable and readable across scenarios.
Directus as the back office for dictionaries, formulas, scenarios, and guidance texts
We deployed Directus as both a headless CMS and an admin panel for HR and operations. It manages dictionaries, coefficients, scenario templates, eligibility rules, and guidance texts. Roles and permissions are separated so HR can update content and parameters, but can't touch the critical technical framework.
AI assistant that explains the forecast and suggests next steps
The AI layer receives the already-calculated result, the current rules, and the dictionary context, and produces a plain-language explanation. It highlights what influenced the forecast, why one scenario performs better, and which specific requirements still aren't met.
AI blocker detector for certifications and requirements
A dedicated AI mode highlights met and missing requirements and shows exactly what's still needed for a given grade or role. It doesn't invent policies — it leans on the rules stored in Directus and presents the gaps in language an employee can actually act on.
AI roadmap for the next 3–12 months
AI generates an action plan against the current state and the user's goal — training, research participation, documentation requirements, quality obligations, audit preparation. The roadmap comes out as a sequence of prioritized steps with dependencies, ready to use.
AI control through Directus
Directus holds the AI assets too — response templates, the center's terminology, recommendation limits, standardized explanation formats. HR can edit phrasing and answer structure without a release, and operations stays in control of what AI is allowed to say.
Audit trail for critical configurations and AI settings
Change history is enabled for the critical entities — coefficients, eligibility rules, formula versions, AI templates. The team can see who changed what and when, and roll back quickly when something goes sideways, without engineering involvement.
We started from the rules themselves — collecting formulas, variables, exceptions, and explanations from HR, compliance, and department leads. Part of the logic lived inside conditional statements like "calculate only if the certification is valid" or "use a different coefficient at this grade." At this stage we aligned measurement units, rounding rules, acceptable input ranges, and behavior for incomplete data, so the system wouldn't produce jumps later because of small interpretation differences.
+2 Resource
After launch, the center has a working internal tool that takes manual calculations off the table and makes career planning transparent for employees.
consolidated into one managed pipeline with explicit dependencies and consistent rounding rules
in daily use for scenario modeling
employee, HR editor, administrator — with real permission boundaries
of results for identical input data, thanks to aligned units, rounding, and a fixed computation order
for a typical AI explanation after a scenario recalculation
generated against the user's goal and the center's rules
on critical entities and AI templates
Or send us a message and we'll get back to you within 15 minutes during business hours