Onboarding
From signature to first working day: accounts, file, badge and starter package that appear by themselves.
Live at pilot client
Typing the same details into six systems for every new colleague.
The problemName, role and start date were retyped six times: time registration, Microsoft 365, SharePoint, badge, telecom provider and the external IT partner. Nobody had one place to see what was already done. The owner put it plainly: "I do not want to think about whether something was done or not."
What we builtOne form in which HR enters the new employee once. That entry fans out in parallel to time registration, Microsoft 365 (account, mailbox, licence), the SharePoint file folder, the badge, plus e-mails to telecom and the IT partner. Every step has its own status and can be re-fired on its own without repeating the rest.
What it deliversOne moment of entry instead of six. The question "has that been done yet?" disappears because the status board shows it. Before the build we estimated about 90 minutes of manual work per onboarding, spread over HR, IT and management (estimate, not measured).
What this becomes for youA hire wizard that fans out to 3 to 6 of your own systems, with each step visibly succeeded or not.
Live at pilot client
What a new employee gets depends on who happens to remember.
The problemWhat exactly a starter received in equipment and access (laptop, phone, badge, software licence) was written down nowhere. It lived in the head of HR or the IT partner and changed with whoever did the hire. Result: forgotten equipment, licences that were too expensive, or access that did not fit the role.
What we builtA profile code made of four choices (department, entity, role, contract) automatically determines the starter package: hardware, badge access, Microsoft licence and the onboarding checklist. HR picks the four parts, the platform derives the rest and includes it in the dispatch.
What it deliversThe package per starter is a row in a table instead of a conversation. Deviations become a deliberate choice instead of an oversight, and HR adjusts the rules itself without code.
What this becomes for youA profile table that derives the starter package and checklist per job profile, linked to your hiring flow.
Whoever changes teams keeps their old access and roles.
The problemAt every team change, roles in time registration, the HR app and the appraisal platform had to be adjusted separately. In practice the old role lingered: someone who had been in another team for months could still see their previous team's data.
What we builtOne organisation code from payroll decides in every tool who may see what. Daily jobs calculate the roles per user, generate the removal import for old rights automatically and report unmapped teams to the administrators.
What it deliversAccess follows the organisation instead of the other way round. A team change in payroll is a team change in every tool the next morning, logged with date and decision-maker. Access management rarely fails at granting and almost always at removing.
What this becomes for youA daily sync of department and role from your payroll provider to Microsoft 365 groups and portal roles, with automatic clean-up on change.
Sound familiar? In a free 30 to 45 minute HR Quickscan we look at where your team loses time here and what should come first.
Book your free HR Quickscan
Leave
Balances, requests and leave pressure straight from the source, without HR keeping track.
At year end nobody knows how much compensatory leave is still open.
The problem"How much leave does my department have left?" meant an export from time registration, a pivot table and half a day of work. Team leads got the answer too late to steer, and nobody saw which balances would expire on their own within weeks.
What we builtAn assistant with dozens of certified questions on current balances: balance per team, who is almost at zero, what expires soon, this year versus last year. The calculation rules (cumulative versus delta, school year versus calendar year, which counters are real leave) are captured once, correctly.
What it deliversHR asks the question and gets the number that is also on the dashboard. The October conversation with each team lead starts with the table instead of with building it. Proactive steering becomes feasible because it no longer costs work.
What this becomes for youA leave balance overview per team from your payroll provider or time registration, with an "almost used up" list and the carry-over rules built in.
Leave requests land with the wrong approver or simply sit there.
The problemThe approval e-mail went to the old manager or a forgotten stand-in, and the request got stuck. Other requests were submitted but never registered as a code, and so fell between two systems and out of the pay run.
What we builtA daily job determines the right approval address per employee, including stand-ins. A weekly check finds every request from the past eight days without a registered code and mails the list with names and departments to payroll.
What it deliversApprovals reach the person who should handle them. Whatever is still unprocessed after a week surfaces at payroll instead of on the payslip. Two small jobs, together the difference between "requests disappear" and "requests get followed up".
What this becomes for youCorrect-approver resolution plus a weekly list of open requests to HR or payroll.
HR checks part-time and time-credit requests against the rules by hand.
The problemParental leave, time credit, progressive return to work: every request had to be checked for the certificate, duration, a valid work schedule and approval status. Errors only surfaced at payroll or at the unemployment office, and applicants waited days without knowing why.
What we builtEvery request is automatically checked against five rules (certificate present, permitted duration, valid schedule, periods, status). What passes goes through; what fails gets a bilingual e-mail with the reason. A waiting-time counter shows where requests get stuck.
What it deliversThe rules are applied consistently, regardless of who reads the request. Applicants get an answer with reasons the same day, and waiting times are measurable. Legal rules around part-time work are exactly the kind of knowledge you do not want living in people's heads.
What this becomes for youRule checks on part-time and time-credit requests with an automatic approval or rejection e-mail and a waiting-time counter.
Sound familiar? In a free 30 to 45 minute HR Quickscan we look at where your team loses time here and what should come first.
Book your free HR Quickscan
Sickness and absence
Sick notes that check themselves, legal deadlines that are no longer missed, and absence you see before it becomes a problem.
Employees do not know whether their sick note arrived, so HR phones everyone individually.
The problemAfter submitting, the employee heard nothing. For a missing or rejected note HR had to check two systems and then phone or e-mail, in the right language. That happened too late or not at all, with pay consequences for the employee and arguments afterwards.
What we builtEvery day at 4 pm a job checks whether each processed note matches a registered sickness code and sends every employee a message in their own language: "sick note in order" or "please send a sick note". HR gets a summary.
What it deliversThe employee knows where they stand the same day. HR no longer chases sick notes, and the "I did not know it was missing" argument disappears. The cheapest automation in absence is not reading the note but the feedback on it.
What this becomes for youA daily automatic confirmation or reminder per employee after the sick-note check, multilingual.
Long-term sick: who needs which letter when, and who forgot?
The problemSickness periods consist of day codes with gaps, weekends and public holidays in them. "How long has someone been ill" is therefore one of the hardest questions in HR data. The weekly letter list hung on a chain of three separate tools; if one link failed, nobody got a letter and nobody noticed.
What we builtA daily job calculates the running uninterrupted sickness period per employee. Every Wednesday a second job builds the mail-ready list of who needs a follow-up letter (address, certificate window, priority) and sends it. The status is written back to the other systems so everything says the same thing.
What it deliversNo legal follow-up deadline is missed anymore. The list is ready every week without anyone building it, and "long-term sick" means the same thing in every system. Get that calculation right once and you can build ten follow-up processes on it.
What this becomes for youCalculate the running sickness period per employee, detect the legal follow-up moments and prepare the letter or e-mail automatically.
We only notice frequent short absence when it is too late.
The problemSomeone who is ill for one day twelve times does not stand out in any monthly report: the sickness rate stays low. HR managers know the Bradford factor, but nobody calculates it. Care conversations were scheduled on gut feeling, sometimes stacked, sometimes never.
What we builtAbsence data is summarised daily per employee and per month, with the Bradford factor as a fixed business rule in one place. Above the threshold the system creates a follow-up conversation automatically, with a minimum interval between conversations and a log of what was done.
What it deliversFrequent short absence is flagged instead of noticed by chance. Conversations follow a rule instead of a feeling, and the Belgian reintegration duty gets a trigger instead of a good intention.
What this becomes for youBradford and frequency thresholds that automatically create a conversation task, with a conversation form and a log.
Sound familiar? In a free 30 to 45 minute HR Quickscan we look at where your team loses time here and what should come first.
Book your free HR Quickscan
Payroll
The checks before the payroll run run themselves, and so does the delivery to the payroll provider.
Someone leaves and we have to settle twelve kinds of balances.
The problemAt every departure HR has to convert all open leave balances into payout or transfer codes, and every counter has its own rules. It existed as a report flow of 166 steps plus manual calculation per leaver. With high turnover that is every week again.
What we builtAn evening job converts twelve kinds of balances to the right codes for every leaver, prepares the import files and updates the status fields in the source system.
What it deliversA 166-step calculation is replaced by one job that runs every evening. HR sees which balances are settled without counting itself, and the payroll instruction is always complete. When a calculation needs 166 steps, it is no longer a report but software.
What this becomes for youOn departure, automatically collect the open balances and generate the payroll instruction, as a step in your offboarding.
Retyping the mileage allowances for the payroll provider every month.
The problemEvery month the bike and car kilometres per employee had to be pulled from one system, put in the right format and brought to the payroll platform. Corrections for previous months were separate manual work and loose e-mails to the payroll provider.
What we builtA job calculates the allowance per employee with monthly eligibility, builds the file in the payroll provider's fixed format, delivers it over SFTP and archives a copy. A four-month correction window carries late changes along automatically.
What it deliversNobody retypes anything and nobody has to start a script. Corrections travel along by themselves. A link with the payroll provider does not have to be an API: a correct file with archive and error alert is in practice more reliable than most "integrations".
What this becomes for youBike and car allowances or other variable pay components delivered automatically as an import file to SD Worx, Acerta or Liantis.
A public holiday fell between two sick days and nobody had seen it.
The problemA paid public holiday between two sick days counts as a sick day in Belgium and falls under a different pay regime. To find it, someone has to lay calendars side by side for hundreds of employees. That does not happen every month, and when it does, manual work misses cases.
What we builtA scheduled check automatically finds every "sandwich" case in the day codes, runs integrity checks and prepares the cases to correct for payroll.
What it deliversThe check runs systematically on the whole population instead of on a sample. Payroll gets a list, not a search. The rules that cost the most manual work are rarely complicated; they are just tedious to apply consistently to hundreds of people.
What this becomes for youNightly checks on the export from your payroll provider or time registration, with a list of deviations by e-mail before the payroll run.
Sound familiar? In a free 30 to 45 minute HR Quickscan we look at where your team loses time here and what should come first.
Book your free HR Quickscan
Contracts
Every date you must not miss, watched by a daily scan.
A temp's contract is ending and the decision meeting is not scheduled yet.
The problemFixed-term contracts need a decision at a fixed moment: extend, make permanent or stop. If that conversation comes too late, the calendar decides instead of the manager. One missed end date costs a severance payment.
What we builtTriggers from the contract data create the conversation themselves: first evaluation, quarterly, and the decision meeting well before the end of the fixed-term contract, with the contract history attached. After the evaluation the contract decision follows automatically with mandatory fields.
What it deliversThe decision moment comes from the contract itself, not from someone's diary. The manager gets the form before the deadline. A daily scan costs nothing; a silent extension does.
What this becomes for youReminders 60 and 30 days before the end of every fixed-term contract, a counter for the rule of four consecutive fixed-term contracts, and the conversation to go with it.
Which certificates does this employee hold, and are they still valid?
The problemCertificates and qualifications decide who may work where. The data sat in the planning system, but "which skills does X have, at which level, valid until when" needed an export. At department level it was pivot-table work, and an expired certificate only surfaced once someone was already scheduled.
What we builtThe qualification tables are mirrored nightly and can be queried in plain language: skills per person with level and validity, skills per department and how many people hold them, and where a whole department depends on one person.
What it deliversThe skills matrix is a question instead of an export. A skill is not a yes or no but a level with an expiry date; once you carry those two fields, the matrix becomes a compliance instrument instead of a list.
What this becomes for youA certificate tracker (safety certificates, first aid, driving licences, forklift) with expiry dates, advance reminders and a compliance meter per department.
Sound familiar? In a free 30 to 45 minute HR Quickscan we look at where your team loses time here and what should come first.
Book your free HR Quickscan
Appraisals
Evaluations, meeting preparation and team overviews that schedule themselves.
Who has not had an appraisal yet this year?
The problemAn appraisal cycle is forms that have to be created, filled in, reminded and closed on time. Without an engine, forms stay open, are not closed after a contract change, and nobody knows which department is behind. Follow-up was list work.
What we builtA 28-step nightly process runs the cycle itself: creating forms based on contract data (end of probation, quarterly, yearly), reminders per role, automatic expiry, closing on contract change and yearly anonymisation. HR queries the completion rate per department.
What it deliversThe cycle runs without anyone keeping a list. HR steers on completion rate and open conversations instead of on memory. An appraisal cycle never fails on the form, always on the follow-up.
What this becomes for youAn appraisal cycle on contract data with automatically created forms, reminders per role and a "who has not had a conversation yet" overview per team.
I have a performance review tomorrow and the information sits in four systems.
The problemA manager preparing has to look in four systems or ask HR for a summary that costs time. In practice the preparation does not happen, and the conversation is about impressions.
What we builtOne question builds a file from four sources: absence and activity, leave and balances, previous evaluations, and open or closed HR questions. Deliberately one fixed, tested composition instead of a router hoping four parts cooperate correctly.
What it deliversPreparation is one question instead of four exports. The conversation starts from facts the employee can see too, and the manager walks in prepared without burdening HR.
What this becomes for youMeeting preparation per employee that bundles absence, leave balance, previous conversations and open HR questions from your payroll provider and your own platform.
Every manager asks HR for the same overview of their team every month.
The problemManagers want to know what is going on in their team, but HR cannot compile a separate overview for everyone every month. So nobody gets anything, or everyone gets the same generic report that nobody reads.
What we builtA monthly job sends every manager one personalised, bilingual e-mail with only the follow-up actions of their own team, with an Excel attachment. Before go-live everything runs in dry-run mode so no e-mail leaves wrongly.
What it deliversEvery manager gets only their own team, automatically, every month. HR no longer compiles overviews. Personalising per recipient is free for a job and unaffordable for a person; that is why a digest gets read and a generic report does not.
What this becomes for youA monthly digest per manager: absence, open conversations, expiring certificates and contracts of their own team.
Sound familiar? In a free 30 to 45 minute HR Quickscan we look at where your team loses time here and what should come first.
Book your free HR Quickscan
Leavers
Notice periods, turnover and departures, calculated in advance instead of requested.
Do people leave, or do we let them go? Nobody could show it.
The problemTurnover sounds simple, but needs exits divided by average headcount, per year, per job level or shift, and split by who took the initiative. Every management team wants that story, and almost nobody gets it out of their systems without days of export work. Three meetings, three turnover figures.
What we builtThe turnover formula and the scope agreements (who counts as an exit, who is in the denominator) are fixed once in a shared rules file. Dashboard and assistant read the same formula: exit initiator per year, exit reasons with percentages, departures per team per month.
What it deliversThe voluntary-versus-involuntary story has become a chat question instead of a quarterly project, with a trend over the years and the same number everywhere.
What this becomes for youTurnover, exit initiator and exit reasons per year and per team from your payroll provider's leaver data, as a monthly report or a chat question.
How many people will we lose by year end? Without showing names.
The problemTeam leads wanted to see departures ahead for capacity planning, but were not allowed to see individual data. A forecast straight on the personnel table gives their assistant access to rows with names and exit reasons. That is a privacy problem, not a filter problem.
What we builtA daily job calculates a small table without personal data: departures per month, daily average, yearly projection. The team-lead version reads only that table and answers "expected departures in 30, 60, 90 days and by year end". The table with names simply does not exist in their version.
What it deliversManagers get a departure forecast without ever seeing a name; HR keeps the detailed version. Privacy is technology that enforces it, not a prompt you have to trust.
What this becomes for youA simple departure projection as a table without personal data for management and team leads. At 50 to 250 employees, read it as a direction, not a prediction.
Sound familiar? In a free 30 to 45 minute HR Quickscan we look at where your team loses time here and what should come first.
Book your free HR Quickscan
Files and figures
One truth per employee, and questions in plain language with the same number as in the report.
Live at pilot client
All HR data is scattered, and every question starts with an export.
The problemEvery system had its own version of the employee: time registration, payroll at the provider, appraisal platform, insurance, tickets, documents. Every report started with exporting and copy-pasting. The most critical monthly report was a manual spreadsheet with lookups, called "a major source of errors" by the organisation itself.
What we builtOne data layer into which every source system is mirrored nightly, and on top one row per employee with 150+ columns plus daily answer tables for headcount, absence, leave pressure and contract mix. Everything that follows (checks, e-mails, AI, dashboards) reads only that layer.
What it deliversChat, dashboard and e-mail give the same number because they read the same table. Lay a clean data layer first and every automation after it is built in days. The same data layer sits under the platform at our pilot client.
What this becomes for youAn HR data hub: a nightly mirror of payroll provider, time registration and HR software in one layer, with the employee table as the first deliverable.
How many people did we have in March, and how many of them actually work?
The problemHeadcount and capacity are two different numbers. Counting heads is easy; how much FTE you lose to part-time, time credit and suspension shows in no payroll tool. Every management question about staffing cost half a day of spreadsheets.
What we builtPre-calculated tables ready every night: contract and FTE situation per day, team history, headcount and contract-mix cubes. On those the assistant answers "headcount this month versus last year", "net FTE per department per contract type" and "FTE lost to part-time and suspension".
What it deliversStaffing figures without spreadsheets, and heads and real capacity side by side. "How many people do I really have available next month?" becomes a question of seconds, with exactly the dashboard number.
What this becomes for youA nightly sync from your payroll provider, one employee table and three to four answer tables (headcount, FTE, contract mix, part-time impact) as the basis for reports and chat.
How do I set that up again? Where does the policy say that?
The problemManuals and policies exist, but nobody reads them. Process questions keep landing with the same two colleagues. And a generic chatbot invents an answer that sounds plausible and is not in the document.
What we builtAn assistant that answers only from your own documents, indexed by section, with mandatory source references down to section level. Deliberately without a vector database. If it is not in there, it says so.
What it deliversThe document became the answer instead of the file nobody opens. Every statement is traceable to a section, and the two colleagues who knew everything get their time back.
What this becomes for youChat with your work regulations, car policy, leave rules and onboarding guide, with mandatory citations and "not in the document" as an honest answer.
Sound familiar? In a free 30 to 45 minute HR Quickscan we look at where your team loses time here and what should come first.
Book your free HR Quickscan
Trustworthy AI
Three design rules built into every IDUS solution. No extra cost, but the answer to "can the AI not make mistakes?"
Can the AI not make mistakes with our numbers?
The problemA language model that writes its own calculations makes exactly the mistakes an analyst spends years unlearning: double counting, including the current month in a yearly average, adding balances across several snapshots. The answer looks perfect and is wrong. And when chat and dashboard differ, nobody believes either anymore.
What we builtThe AI may phrase, not calculate. Every frequently asked question lands on a pre-approved, tested calculation that gives exactly the same number as the dashboard. If it finds no data, it says so honestly instead of inventing something. Every rule is under version control.
What it deliversThe assistant gives exactly the number from the report, or says it does not know. Reliability comes from architecture, not from a smarter model.
What this becomes for youStandard in every assistant or reporting layer we deliver.
Live at pilot client
Team leads want figures, but may not see files.
The problemThe classic solution is one chatbot with the instruction "do not show names to managers". That is policy you have to trust. One clever question and it leaks, and with sickness or exit data that is a GDPR incident.
What we builtTwo windows on the same data. HR sees files; managers get separate, aggregated tables without personal data. Their version cannot show names, because those tables do not exist in its reach. Before every AI call, personal data is stripped in two passes; the national ID number is never stored.
What it deliversPrivacy is technology that cannot do otherwise, not a promise. "The team lead cannot see it" convinces a data protection officer faster than ten pages of policy. Enforced at our pilot client with row-level security and field encryption.
What this becomes for youOne role model with separate views for management and team leads, enforced in the database, not in the prompt. Standard in every IDUS platform.
The AI assistant said something wrong, and HR believed it.
The problemOne invented answer about an insurance premium or a leave rule costs more trust than a hundred correct answers build. Trust in an HR assistant is not "mostly right" but "never invented".
What we builtFive layers that teach an assistant to stay silent: hard rules without outside knowledge, citing only from retrieved sections, a hard gate (no source, no answer), post-checks, and a confidence marker on every answer. Every error that appeared in production became an explicit prohibition in the rules.
What it deliversThe system would rather say "I do not know" than something that is not in the source. Usage, cost and thumbs up or down are logged per user, so you know what the assistant delivers and where it still fails.
What this becomes for youThe same five layers plus a monthly usage report, standard in every chat solution.
Sound familiar? In a free 30 to 45 minute HR Quickscan we look at where your team loses time here and what should come first.
Book your free HR Quickscan
Every case above was built for a specific organisation. What we build for you, we decide together in the free HR Quickscan: first see where your team loses time, then choose what comes first.