Cookie preferences
Essential cookies keep this site working. With your permission, we use cookies to recognise you when you return and to remember your activity history. External content such as Google services may also be loaded. You can withdraw your consent at any time.
Project
Competency Measurement and Development Platform
An end-to-end platform developed for an organization with thousands of employees that brings AI-powered exam generation, psychometric measurement, competency analytics, and training management together under one roof.
JustApt is an AI-powered competency measurement and development platform built for a large-scale organization. It measures employees’ technical and behavioral competencies, shows the gap between their level and the level their role expects, and automatically builds the development path that will close that gap.
The platform brings four engines together under one roof: AI-powered exam generation, psychometric measurement, competency analytics, and learning management (LMS). Managers, instructors, and employees work on the same data model, but each sees an interface focused only on their own work. The system connects to the organization’s HR, identity, and organization systems through APIs. The architecture was built with KVKK (the Turkish Personal Data Protection Law) and GDPR requirements in mind.
The organization’s HR team assessed hundreds of employees every quarter, but there was no system built for it. Questions were written by hand by expert teams, matched to competencies by hand, and the results were scattered across different spreadsheets. Seeing a department’s real competency gap was a compilation job that took weeks. Decisions about which training to assign to whom were also usually based not on data but on a manager’s observation.
The request that came to Karatsoft was initially for a learning management system. During the analysis phase, we saw that the real problem was not training content but measurement. A competency that cannot be measured cannot be matched with the right training. So we widened the scope and designed an end-to-end platform that runs from measurement to development: generate the exam, administer it, analyze it, find the gap, assign the development path, track progress.
The resulting product is a system in which a manager generates a role-based assessment with a one-sentence prompt, the results automatically feed into the skills matrix and career steps, and employees see on their own screen what is still missing for the next level.
Tell us what you need; we'll set out the scope, integrations and roadmap in writing in the first meetings. Built around the way you work, without touching your ERP, accounting or banking systems.
The main screens and their highlights: on every image we magnify the area that matters most. Tap an image for details.
Managers and the HR team see the organization’s competency status on a single screen: the cards at the top show the strongest competency, the priority development area, participation, and the completion rate. The competency map compares the current profile with the target profile at the employee, team, and organization level; the 12-month trend, the recommended development program, and the most recently assigned assessments are on the same page.
When a manager selects the department and role, target levels are loaded into the matrix; each employee’s level in each competency sits next to the role target. A gap is measured not by color but by level difference, and differences of two or more levels are flagged separately. The AI panel on the right highlights the critical gap, high-potential employees, and signs of decline.
In a five-step wizard, the manager selects the assignment type, security level, and target audience, then writes what they want to measure in their own words. The system detects the relevant competencies and target level from the text and presents them together with settings for question count, difficulty, duration, and question type. Questions are generated to match the organization’s competency dictionary and role targets.
Before publication, the questions generated by AI come before the expert on this screen. In the question explorer, each question opens with its competency, level, and difficulty; the correct answer is shown together with its rationale. The expert edits a question they are not happy with using a prompt in natural language, has it regenerated, or removes it; the right-hand panel reports quality control and the difficulty distribution.
Employees answer the exam on a clean screen: a countdown timer and a progress bar at the top, and on the right a question map showing answered and flagged questions. In proctored mode, the camera, microphone, and full-screen status appear in the proctoring panel, and tab switching is logged. Every answer is written to the server instantly; no answer is lost even if the connection drops.
Alongside knowledge exams, a personality inventory, psychological resilience, abstract reasoning, and a general ability battery run in the same center. Each scale is calculated with its own scoring algorithm; the results are combined in a single candidate report with the personality profile, resilience sub-dimensions, and cognitive ability indicators. The report is published after approval by the subject-matter expert.
Assessment results are matched to career steps automatically. For each competency, the career progression matrix places the level a step expects next to the person’s current level; how many levels are missing for the next step is visible row by row. Promotion readiness and the recommended development path are on the same page; employees also see this information in their own area.
After measurement, development continues on the same platform. Instructors and the HR team manage the course catalog, classes, assignments, announcements, and badges from a single center; each course’s class count, learner count, and publishing status are visible in the catalog. The monthly calendar separates lessons, assignment submissions, and exams by color; bulk notifications go out with a single button.
The learner area is where employees take ownership of their own development: on opening, the lesson they left off at appears, and a single button takes them back to it. Course cards show the completion percentage and the next unit; upcoming live lessons, assignment submissions, and assessments are listed in the side column. Badges, points, and streaks support learning continuity, and reports show the weekly performance trend.
Requests from the manager, instructor, and employee interfaces reach the API gateway over an encrypted channel; the gateway handles authentication, role-based authorization, rate limiting, and audit logging. The exam, psychometrics, analytics, and training capabilities run as independent services and scale piece by piece. Data is stored encrypted; non-production environments work only with masked data.
The manager writes what they want to measure in their own words; questions matched to a competency and level are ready in seconds.
Every question is reviewed, edited with a prompt, or regenerated; no question goes live without approval.
Checks for competency distribution, difficulty balance, a single correct answer, and duplicates are reported for every set.
One shared scale is used across the entire organization, from the “Needs development” level to the “Can coach” level.
When a department and role are selected, target levels are loaded and gaps appear as level differences.
Critical gaps, high-potential employees, and signs of decline are flagged automatically.
Flexible mode is used for development-oriented measurements; proctoring with camera, microphone, and tab tracking is used for critical decisions.
Every answer is written to the server instantly; no data is lost even if the connection drops.
A personality inventory, psychological resilience, abstract reasoning, and a general ability battery run on the same platform.
Behavioral tendencies are numerically compared with the profile the organization expects.
Competency expectations are defined for each career step and compared with the person’s current level.
An estimated readiness time is calculated based on the number of gaps and the recommended path.
Training, practice, and mentoring steps are selected from the organization’s training catalog and assigned to the employee.
Courses, classes, assignments, calendar, announcements, messaging, and newsletters are managed from a single center.
Badges, points, and streaks support learning continuity.
Managers, instructors, and employees see only the data they are authorized to see; the principle of least privilege is applied.
Only masked data is used in all non-production environments.
A two-way, synchronous data flow is provided with HR, identity, and ERP/CRM systems.
Offers an infrastructure that is WCAG 2.1 AA compliant, screen-reader friendly, and ready for multiple languages.
Keep scrolling
Move between slides with the arrow keys or by swiping.
How we work
This project was run on the same four principles; yours will be too.
Scope, integrations and roadmap are written down and agreed in the first weeks; no surprises along the way.
ERP, accounting, banking, e-invoicing: whatever you run, we integrate on top of it and your data stays where it is.
Role-based access, two-factor authentication, data-protection compliance and a log of every action are built into the foundation, not bolted on later.
Maintenance, monitoring and new releases come from the team that built the project; someone who knows your system is always a call away.
JustApt
The manager cockpit gathers the organization’s competency status and the progress of assigned assessments on a single page. Managers see the strong and weak areas, the deviation from the target profile, and the next development step without compiling separate reports.
The four cards at the top show the organization’s strongest competency, its priority development area, participation, and the completion rate against target. The competency map overlays the current profile on the target profile; the manager switches the view to the employee, team, or organization level. The development chart tracks the monthly average score over twelve months. For an area that shows a decline, the system recommends a development program, and the manager assigns the program from the same card. The table below lists the latest assessments with their target audience and completion rate.
Seeing a department’s real competency gap used to mean weeks of compiling results scattered across different spreadsheets. The cockpit draws on the same data model as exams, the skills matrix, and the training modules, so the overall picture is available at any moment. Managers determine which area is declining and which program should go to whom by looking at data, not at observation. When a new measurement is needed, the buttons for assigning an exam and for creating one with AI sit in the top bar of the screen.
The skills matrix places the competency level of everyone in a role next to the target the role expects. The manager sees in a single table who is how many levels behind in which competency, and which gap needs to be closed first.
When a department and role are selected, each competency’s target level is loaded into the column header. The cells use the common five-level model: needs development, beginner, basic, proficient, and can coach. Cells that fall two or more levels below the target carry a separate marker, and the fit percentage at the end of each row shows how fully the employee meets the role target. The cards at the top summarize the current competency, the skill gap, and the most critical gap.
When departments used the same competency under different names and with different level definitions, comparing teams was impossible. Thanks to the central competency dictionary, the matrix, the exams, and the career steps speak the same language. The insights panel automatically flags the critical gap, high-potential employees who could act as mentors, and people whose latest assessments show a decline. The manager turns a recommended action into a development path with one button; the training decision rests on a measured level difference.
Exam preparation goes from a task that takes expert teams weeks down to a single sentence written by the manager. The wizard matches that sentence against the competency dictionary and generates a question set suited to the role target.
The wizard has five steps: assignment type, security, target audience, prompt, and confirmation. In the prompt step, the manager writes the goal in their own words. The system scans the competency dictionary, the role targets, and the question pool, and shows the competencies and target level it detects in the text as labels. The manager sets the question count, difficulty, duration, and question type in the same step. Two options make it possible to use the organization’s dictionary and role targets and to exclude similar questions asked in the last six months.
In the past, expert teams wrote questions by hand and matched them to competencies one by one; preparation for each role took weeks. The wizard prepares the question set in seconds but does not let go of control: every question comes tied to a competency and a level, and goes through expert review before publication. The HR team gets faster, the reliability of the assessment is preserved, and the questions stay consistent with the organization’s shared competency language.
The question preview screen combines the speed of AI with expert control. The expert reviews every question one by one, and no question goes live without approval. Why each question is correct and which level it measures stays on record.
The explorer on the left lists all the questions in the set with a competency label and a difficulty indicator. The selected question opens in the middle with its competency, level, and difficulty labels; the correct option is marked, and its rationale is written below. An instruction typed into the edit-by-prompt box updates the options, the difficulty, or the context. The right-hand panel shows how many questions were generated from each competency and the five-step difficulty distribution; it flags the checks for a single correct answer, duplicates, and competency match.
A single wrong or ambiguous question generated automatically can make the entire assessment disputable. While preserving speed, this screen keeps expert control at the center of the process: the subject-matter expert sees every question, reads its rationale, and, if necessary, has it corrected with a single sentence. Because quality control reports the balance of the set automatically, the expert spends time on the content rather than on counting. When the set is ready, the expert saves it as a draft or publishes it and assigns it directly.
The proctored exam screen gives the employee a clean space to focus in and gives the organization a reliable result. The security level is chosen per exam; development measurements run in flexible mode, while promotion and hiring assessments run in proctored mode.
The exam name, an indication that proctoring is on, and the remaining time appear in the top bar. The question card carries the competency and level label; the answer options sit under the scenario text. An employee can flag a question to come back to later; the question map shows answered, flagged, and empty questions in separate colors. The proctoring panel monitors the status of the camera, microphone, full screen, and connection. Tab switching and leaving the screen are logged; when time runs out, the answers are submitted automatically.
Results used in critical decisions such as promotions and hiring must be protected against cheating and outside help; in development-oriented measurements, the same strictness creates unnecessary friction. The organization sets this balance separately for each exam: flexible mode, or proctored mode with camera, microphone, full-screen, and tab tracking. Because answers are written to the server instantly, a connection problem does not waste the employee’s effort.
The psychometric measurement center runs personality, resilience, and cognitive ability scales on the same platform as knowledge exams. The results of the four scales, together with an organizational culture fit score, turn into a single expert-approved candidate report.
The personality inventory addresses sixteen core personality traits individually; psychological resilience measures the dimensions of adaptability, self-control, self-efficacy, optimism, and determination. Abstract reasoning works with questions made of visual patterns that do not depend on language. The general ability battery covers sub-areas such as numerical ability, making connections, and synthesis. In the report, the personality profile is shown on bipolar scales, resilience with bars, and cognitive ability as a percentile against the organization’s norm group.
The scoring logic of personality, resilience, and cognitive ability scales is completely different from that of knowledge exams; even so, they all had to be combined consistently in the same report. The platform scores the four scales with separate algorithms and gathers the results in a single candidate report. Behavioral tendencies are numerically compared with the profile the organization expects, and a culture fit score is produced. The final report is not published until the subject-matter expert approves it; hiring and promotion decisions rest on data that has passed expert review.
The individual development map turns measurement into action: on a single page it shows in which competency a person falls short for the next career step and by how much, how ready they are, and which steps will close the gap.
The career progression matrix defines the level expected in each competency for each step, on a scale of one to five. The person’s current step and the next step are highlighted; each row states the number of missing levels or that the person is ready. A radar chart compares the current level with the expectation of the next level. The promotion readiness indicator calculates an estimated readiness time from the number of gaps and the recommended path. The development path lists training, practice, and mentoring steps with their progress status.
A competency that cannot be measured cannot be matched with the right training; this screen ties the measurement result directly to a development step. The employee sees what is missing for the next level and how to close it, and the manager bases the promotion decision on the expectations of the steps. Because the path steps are chosen from the organization’s training catalog, the recommendation stays actionable; HR edits the path if needed. The estimated readiness time helps the HR team set the promotion plan on a realistic timeline.
Training management runs the learning process that will close the gap found through measurement on the same platform. Courses, classes, assignments, calendar, announcements, messaging, and newsletters are managed from a single center; there is no need to move data to a separate learning system.
Tabs let users move between courses, classes, assignments, announcements, and badges. The cards at the top summarize the number of active courses, enrolled learners, and pending assignments, and the average completion rate together with its target. The course catalog shows each course’s instructor, number of classes, number of learners, and whether it is published or in draft; a new course is opened from the same panel. The calendar gathers live lessons, assignment due dates, and exams in a single role-based view.
The request that came to Karatsoft was initially for a learning management system; when the analysis showed that the real problem was measurement, training management took its place as the last step of the flow that runs from measurement to development. The training steps in the development path are selected from this catalog; the path closes the measured gap with the organization’s own courses. The instructor tracks learner progress instantly; bulk notifications, in-system messaging, newsletters, and announcements keep communication in one place instead of scattering it across separate tools.
The learner area gives the employee their own development journey on a single page: they continue where they left off, see upcoming tasks, and earn badges and points as they progress. A reinforcement module recommended from their latest assessment sits on the same page.
The page opens with a personal welcome and a weekly learning goal. The “Continue where you left off” card shows the last lesson opened, the course it belongs to, and the lesson type. In the “My courses” section, each course card carries the completion percentage and the next unit; the catalog is one link away. The “My reports” chart plots the weekly trend of assessment performance. The “Upcoming” list orders live lessons, assignment submissions, and assessments by date; the development recommendation card starts the module directly.
A development path only works if the employee keeps it going. The learner area brings the path into the employee’s everyday flow: live lessons, assignments, and assessments appear in a single list, so due dates are not scattered across separate conversations. Badges, points, and streaks reward steady progress. The employee sees in which competency they fell behind in their latest assessment and starts the recommended reinforcement module from the same page; the instructor, for their part, tracks progress instantly.
The platform is based on an API-first architecture that builds every capability as an independent service. Security is handled layer by layer, from communication to the data layer; the design takes KVKK (the Turkish Personal Data Protection Law) and GDPR requirements into account from day one.
The user channels consist of the manager panel, the instructor panel, the employee and learner area, and a mobile-friendly interface. Traffic reaches the API gateway over an encrypted channel; the gateway handles authentication and optional two-factor authentication, role-based authorization, rate limiting, and audit logging. Behind it, the exam and question engine, the AI generation service, psychometric scoring, competency analytics, training management, and the notification and calendar services run separately. The data layer provides masking, encrypted storage, environment isolation, and backup.
Assessment results are among the organization’s most sensitive data; for this reason the architecture treats privacy not as a layer added afterward but as part of the design. Development and staging environments are completely separate from production and work only with masked data. The application is protected against the OWASP Top 10 risks. For organization systems that were not yet ready, we defined the integration contracts in advance; when the other side was ready, the connection went live without additional development.
