HR automation without a spreadsheet policing the dates
Leave requests, day balances, health check and training expiry, timesheets and the numbers payroll needs stop circling between inboxes and spreadsheets. The workflow pulls them from the systems you already run and watches the dates itself. HR keeps the part of the job where somebody has to talk to a person.
- 14 daysfrom the workshop to the first working HR automation, usually leave requests or the expiry alerts on health checks and training
- 30 dayshow far ahead the alert goes out before a health check, a training record or a licence expires, to the employee and to the manager
- 5 to 10 secondshow long an AI agent takes to answer who is off this week, straight from the leave system and the team calendars
The problem
Why does HR police deadlines in a spreadsheet instead of a system?
Monday in HR looks much the same at 40 people and at 300. Three leave requests in the inbox, two of them with nobody having checked the balance. A manager asking on Slack who from his team is around this week. A candidate from last week still waiting for a reply, because the application is sitting in the recruitment inbox between messages from agencies.
The second layer is made of dates, and that is the layer that turns into a problem at an audit. Occupational health checks expire. Safety training expires. Operator licences and professional certifications expire, per person and per piece of equipment. Carry over leave has a cut-off date. Employee records have a retention period that starts running on a specific day. Every one of those dates is currently watched by somebody's spreadsheet and somebody's memory.
Part of it is not a matter of good practice. GDPR 2016/679 in Art. 5(1)(e) says personal data may be kept only as long as there is a basis for keeping it, which is exactly the rule that applications from candidates you did not hire tend to break, because nobody deletes a folder just in case. The AI Act 2024/1689 puts recruitment and selection of candidates in Annex III, so any tool that ranks or filters applicants sits in the high risk category and needs a person signing off the decision. On top of that sits national labour law, with its own working time records, its own leave entitlement and its own retention periods, different in every country you employ people in. A spreadsheet that holds all of that is a single point of failure with a name and a holiday allowance.
The third layer is the end of the month. Hours from the time records, overtime, absences and changes to contracted hours have to be collected, checked and retyped into payroll. Retyping the same figures between the HR system, a spreadsheet and the payroll tool is where the errors come from, and they only become visible after the transfer has gone out.
- A leave request sits in the inbox because the person who approves it is off
- A health check or a safety training expires and nobody spots it until after the date
- A candidate gets no reply and closes the subject with your competitor
- Payroll data is retyped by hand from three places, once a month, every month
Why now
Can this be done without replacing the HR system?
Yes, and that is the main thing that changed over the last few years. The systems you already run expose APIs: Personio, BambooHR, HiBob, Sage, Calamari, Workday, Google Workspace and Microsoft 365. A leave request can come in through a form or a Slack command, get its balance checked in the HR system, go to the manager for approval and land in the team calendar, with no email to HR at any point. The automation sits between the systems, not instead of them.
The second change is on the document side. A contract amendment, a declaration, a referral for a health check or a consent form goes from a template into DocuSign, Adobe Acrobat Sign or Autenti, and comes back as a file with a date on it, ready to file. The clock then starts from an event in a system rather than from the moment somebody put a signed sheet of paper on a desk.
The third change is the AI agent. A workflow will count a balance and raise an alert, but it will not answer a manager asking who on my team still has carry over leave left. BOND, our company AI agent, answers that in a plain sentence in 5 to 10 seconds, straight from the HR data, with nobody logging in and nobody building a report.
What changes
What does the HR team actually get?
A candidate hears back the same day
Applications from job boards, the careers form on your site and the recruitment inbox land in one queue, each with a date and a source. The candidate knows immediately that the application arrived and what the next stage is, and the recruiter works from a list instead of an inbox.
One candidate queue instead of three channels, each with a different owner
Leave without an email to HR
The employee files through a form or on Slack, the workflow checks the balance and the contracted hours, sends it to the manager for approval and writes the absence into the team calendar once it is signed off. The stand-in finds out at the same moment.
The balance is counted from the HR system, not from a spreadsheet two months old
Deadlines remind you by themselves
Health checks, safety training, licence validity and carry over leave all have a date in a system, so the alert goes out before the deadline and to a named person. Not to a shared HR inbox where everyone assumes somebody else will pick it up.
One alert 30 days before the expiry date and a second 7 days before, to the employee and the manager
Payroll data ready on the first of the month
Hours from the clock in system or from time reports, overtime, absences and changes to contracted hours come together as one set, in the format the payroll tool accepts. HR checks the exceptions instead of retyping everything.
One file for payroll instead of three spreadsheets merged by hand
A leaver closes with one procedure
The end date triggers a list: leave settlement, final documents, equipment return, access revocation, handover of duties and the exit conversation. Every step has an owner and a date, so nothing is left for later.
Access revoked on the last day, not three weeks after somebody notices the account is still live
HR questions answered in a plain sentence
Who is off today. Whose health check expires this quarter. How many carry over days are left in the sales team. An AI agent answers from the HR system and the calendars, without anyone opening a report.
An answer in 5 to 10 seconds, with no login to the HR system
Proof
Do you watch your own team availability this way?
Yes. Availability in our founders' agency is watched by an AI agent, the same one behind the numbers on our AI automation agency page. Below is the scale of that work, together with what the number does NOT prove.
Built for ourselves. These are not client deployments but implementations inside the founders' own companies. We say so plainly, because proof you cannot verify is not proof.
JustBottheir agentJustIdea Agency, a performance marketing agency of around 30 people
- 23
- people from the company work with the agent
- 38
- systems wired in permanently
- 112
- processes run without a human
- 475
- thousand lines of the agent’s code
The agent is wired into the leave system, the team calendars, Slack, the task system and the data warehouse. 23 people use it, the project management team included.
8 of 38 systems it works in
Google Ads
Meta Ads- HubSpot
- ActiveCollab
- Ahrefs
GA4
Search Console
BigQuery
- Before
- Checking who was available meant asking on Slack or opening a spreadsheet. The answer depended on who happened to know who was off and until when.
- After
- The agent answers questions about absence and availability from the leave system and the calendars. A person steps in only where something in the work plan actually has to be moved.
8,631 operations on team availability in 30 days, with no human involved
And in your company
What if our HR works differently?
Four sentences we hear most often, and what we say back to them.
We are 25 people, leave lives in a spreadsheet and it works.
At 25 people it often genuinely does work, and we say so on the consultation. The sensible move then is to tidy up the spreadsheet itself and add one reminder about expiry dates. We come back to it when more than one person starts counting leave, or when that one person leaves the company.
We already have Personio, why would we need anything more?
If the system covers the whole process, we will tell you that and it ends at the workshop. Automation goes in where the data has to travel onward: into payroll, into the team calendar, into a ticket for IT, into a report for the board. Those handoffs between systems are what eats the hours, not the HR system itself.
HR data is the most sensitive data in the company.
Which is why the set of fields is agreed in the workshop and only what a decision needs gets passed over. The data stays in your systems, we do not set up an employee database of our own. We sign a data processing agreement and we set retention periods separately for employees and for candidates you did not hire.
Our HR team has no time for a rollout.
The workshop is one week of work with us, not a project that eats a quarter. The pilot runs alongside the current way of working, so HR does not have to switch anything off or learn two methods at once. We take the process that costs the most hours first, so the time you put in comes back fastest.
- HR systems
- Personio, BambooHR, HiBob, Sage, Calamari, Workday
- Recruitment
- Greenhouse, Lever, SmartRecruiters, job boards, the careers form on your site, the recruitment inbox
- Documents and e-signature
- DocuSign, Adobe Acrobat Sign, Autenti, Google Workspace, Microsoft 365
- Time records and payroll
- clock in systems, timesheet reports, payroll and ERP exports, Google Sheets
- Team work and data
- Slack, Microsoft Teams, Notion, ClickUp, Jira, Asana, BigQuery
- The automation layer
- n8n, Make, Zapier, custom API integrations
HR deadlines a workflow can watch for you
None of these rows is our list of good practice. Each one is a date that already exists, either in law or in a document somebody signed, and each one decides when an alert has to go out and to whom. The exact figures differ by country, so the workshop fills them in for the countries you employ people in.
| Deadline or limit | Where it comes from | What the workflow does |
|---|---|---|
| Annual leave entitlement per person | the employment contract plus the statutory minimum where the person is employed | counts the balance from the HR system and holds a request with no days behind it |
| Carry over leave cut-off | national labour law, with a different date in every country you employ in | alerts the employee and the manager weeks before the date, while there is still time to book |
| Occupational health check validity | the certificate itself, which carries an expiry date | counts from that expiry date, first alert 30 days before, second 7 days before |
| Safety training validity | the training record and the local renewal period | the same two alerts, to the employee and to whoever books the course |
| Operator licence or certification expiry | the licence document, tracked per person and per piece of equipment | keeps one register with owners and dates instead of a tab in a spreadsheet |
| Employee records retention | national labour law sets the floor, GDPR Art. 5(1)(e) sets the ceiling | watches the document set and the date the retention period starts running from |
| Candidate data retention | GDPR Art. 5(1)(e): data with no basis to keep it has to go | deletes or anonymises an application once the agreed period runs out |
| Human sign-off on a hiring decision | AI Act 2024/1689, Annex III lists recruitment among high risk uses | routes the ranked shortlist to a recruiter and never rejects anyone by itself |
Not every row applies to every company. In the workshop we settle which of these deadlines are a real problem for you, and only those go into the first rollout.
How we work
Five steps from a spreadsheet to a working workflow
Process first, tools second. Every step ends with something you can judge before you decide on the next one.
Workshop
We take the HR processes apart: who files, who approves, where the balance lives, where the deadline comes from and how many times the same data gets retyped. The output is a list of processes ranked by the hours they cost every month.
Picking the first process
For the first rollout we take one process, most often leave requests or expiry alerts on health checks and training, because both have a clear rule and a visible result. We agree the set of personal data, the approval path and how long the data has to be kept.
Pilot on real data
The workflow runs alongside the current way of working, with nothing switched off. We test it on your own requests and deadlines from the last few weeks, including the awkward cases: leave that crosses the year end, a change to contracted hours, an absence in the middle of a settlement period.
Production and training
Once the results check out we switch to production with error monitoring. The HR team learns to change the rules without us: a new absence type, a different approver, a different alert threshold, an extra certification with its own expiry date.
Measurement at 30 and 90 days
We come back with numbers: how many requests went through without HR touching them, how many alerts went out ahead of the deadline, how many hours disappeared from the end of the month. Only after that measurement do we decide which process goes in next.
Before and after
How the week changes in HR
Before
- Leave requests wait until somebody checks the balance in a spreadsheet
- A health check expiry gets noticed a week after the date
- The end of the month means two days of retyping data into payroll
- The candidate from last week is still waiting for any reply at all
After
- The request follows the approval path by itself and the balance comes from the system
- The deadline alert goes out 30 days ahead, to the employee and the manager
- Payroll data is ready on the first of the month and HR checks the exceptions
- The candidate gets a confirmation the same day and the recruiter sees a queue
The first process runs 14 days after the workshop. The effect on HR hours is judged at 30 days and the effect on deadlines at 90, because only then has a full cycle of health checks and training gone through.
For the person signing this off
What an HR director gets out of it
The questions has anyone's health check run out and has anyone lost their carry over leave stop depending on one person's memory. The deadline has a date in a system and the alert has an owner, so an audit becomes a review of documents rather than a reconstruction of history from email threads.
The second thing counts in the conversation with the board. You know how many hours a month the team hands over to administration and which process takes them. That is an argument for a headcount or a tools budget, and it is an argument you do not currently have in numbers.
- Statutory deadlines watched by a system rather than by somebody in HR remembering
- The set of personal data written down together with its retention period
- A pilot running in parallel, so the current way of working holds until testing ends
- Rules changed by the HR team itself, without raising a ticket with us
What it costs
The workshop has a published price, the rollout is quoted after it
The process workshop costs EUR 3,500 net per week and runs from 1 to 7 weeks. HR usually comes to one or two, because there are several processes and they have to be put in order before anything gets built. We quote the rollout after the workshop, once it is clear how many processes go in, which systems have to be joined up and what state the HR data is in.
You pay the cost of running the workflow directly to the tool provider, n8n, Make or Zapier, or you pay nothing at all when it runs on your own infrastructure. Same with the language model once an AI agent goes in, where a single query costs the provider roughly EUR 0.005 to 0.07. There is no margin from us on any of it.
Process workshop
EUR 3,500 net
per week, and with HR that is usually one to two weeks
A self-contained piece of work. It ends with a map of your HR processes, the hours each costs and a rollout order you could execute without us.
- A map of HR processes with the hours they cost today
- A list of the deadlines currently watched by a spreadsheet, and by whom
- A tool recommendation: features of your HR system, n8n, Make or a direct API
- The rollout order plus the metrics to measure at 30 days
- Most popular
Workshop and HR rollout
Quoted after the workshop
scope follows from the number of processes and systems involved
The option most companies choose. The first process runs 14 days after the workshop, each one after that goes in within 2 to 4 weeks.
- Everything in the workshop
- A workflow in n8n, Make or over the API, wired into the HR system and the calendars
- Leave requests, deadline alerts, time records and the data set for payroll
- A pilot on your own data and a switchover with error monitoring
- Training so the HR team can change the rules on its own
HR and an AI agent
Quoted after the workshop
for companies where HR answers the same questions several times a day
Everything in the rollout plus BOND, which answers questions about leave, deadlines and availability in a plain sentence.
- Everything in the HR rollout
- An AI agent wired into the HR system, the calendars and the time records
- Alerts on expiry dates and carry over leave routed to the right people
- Manager access, so they stop asking HR who on their team is around
- Agent rollout takes around 2 months, independently of the workflow
The workshop ends with a document that stands up without us: a map of the HR processes, the hours each one costs today and the order of rollout. If it turns out the features your HR system already has are enough, we say so and show you how to switch them on. The engagement then ends at the workshop, not at a rollout you do not need.
Frequently asked questions
What is HR automation?
It is joining up the HR processes so that data stops being retyped by hand between systems. A leave request checks its own balance and goes for approval, a health check expiry reminds you before the date, hours from the time records assemble into a set for payroll, and a leaver triggers one list of tasks. People decide where a situation needs judging, rather than where a number needs copying.
How long does an HR automation rollout take?
One or two weeks of workshop, then 14 days to the first working process and 2 to 4 weeks for each one after that. We do not build every HR process at once, because the first one surfaces your exceptions and there are always exceptions. An AI agent, if it is in scope, is a separate piece of work of around 2 months.
Do we have to replace our HR system?
No. The automation sits between the systems you already run: Personio, BambooHR, HiBob, Sage, Calamari, Workday, Google Workspace or Microsoft 365. There is one condition, the system has to expose an API or accept a file. If part of the process cannot be pulled out of it, we say so in the workshop instead of promising an integration.
What happens to employee and candidate personal data?
Only the fields a decision needs get passed into the workflow, and that set is agreed in the workshop. HR data stays in your systems, we do not set up an employee database of our own. Retention for applications from candidates you did not hire is agreed separately, because Art. 5(1)(e) of the GDPR requires data to go once there is no longer a basis for keeping it.
Can the automation reject a candidate?
No, and we design it that way on purpose. The workflow gathers applications in one place, fills in the details, sends the confirmation and keeps the stages moving. The decision on a candidate belongs to the recruiter, partly because the AI Act 2024/1689 lists recruitment and candidate selection in Annex III as a high risk use with a human oversight obligation attached.
How does the workflow count a leave balance across different contracts?
From the data in the HR system: the entitlement that applies to that person, the fraction of full time hours, leave carried over from last year and days already taken. Leave types with their own limits, such as carer's leave or emergency days, are tracked as separate lines. Edge cases like leave across a year end or a change of hours mid month get tested on your own data during the pilot.
How is this different from onboarding automation?
Onboarding covers the first days of one person: accounts, access, documents to sign and the checks that have to clear before someone can start work. This page is about the rest of HR, the part that runs continuously: recruitment, leave, time records, recurring expiry dates and leavers. We describe onboarding separately because it has a different trigger and different owners for each step.
What about time records for remote work and overtime?
Recording working time is the employer's obligation regardless of where the work happens, so the workflow collects hours from whatever source you have: a clock in system, reports in the task tool or a simple form. Overtime, on-call time and absences are tagged separately, so payroll receives them in a form the payroll tool accepts without a manual correction.
Let us find which HR process costs you the most hours
The free consultation takes 30 minutes. You tell us how many people you employ, where leave is tracked today and who watches the expiry dates. We tell you what can be joined up straight away and what needs the HR data tidied up first. You do not have to share systems or employee records at this stage.
We run at most two rollouts at a time, so that someone is genuinely sitting on each of them.
Book a free consultationPS. If somebody at your company lost carry over leave this year, or a health check expired before anyone noticed, that is exactly this problem. A deadline watched by one person's memory will eventually fall in the week that person is off. A workflow does not have that week. You write directly to Pawel, no reception and no tickets.