DocsSign inInstall Kastel
All connectors

Factorial can feed your AI's org chart without opening pay or leave to it.

With Kastel

Plug Factorial in and your Kastel reads two lists, employees and teams. From an employee it keeps the name, the job title, the team and the manager, nothing else. Pay, leave, absences and personal contact details are never even requested. And the whole directory waits for an administrator to name the department allowed to read it.

An HR system knows far more than everyone's job title

Factorial tracks leave, absences, expenses, pay and appraisals. An HR lead hesitating to plug an AI into it wants a precise answer about what will come out, field by field. This page is that list.

Kastel's Factorial connector cares about the organisation only. It reads two lists, employees and teams. From each employee it keeps four things, the name, the job title, the team and the manager. Each team becomes a record of its own carrying its name, which gives your AI the structure of the company without telling it anything about your employees' lives.

Four fields of the org chart, nothing from the file

Scope is described here in fields, because an HR system mixes the organisation of the company and the private life of a person inside the same record.

What your Kastel reads in Factorial

  • The employee's full name
  • Their job title, when Factorial holds one
  • Their team, when the link resolves
  • Their manager's name, resolved from an id
  • The list of your teams, one record per team

What it never reads

  • Pay and payroll elements
  • Leave, absences and sick notes
  • Health data
  • Appraisals and performance reviews
  • Bank details, personal contact details, date of birth and national identification number

These exclusions run through the easiest path to verify. The connector does not ask Factorial for any of it. Two addresses are called, and neither holds a leave request, an amount, or a piece of health data.

Two lists called, and no other address

Factorial's interface returns whole records, with no way to ask for a few fields only. The connector therefore applies two barriers instead of one, and the first sits on the network.

The engine's Factorial reader knows two addresses, the employee list and the team list. The parts of Factorial holding pay, bank details, personal contact details and dates of birth are never called. None of those addresses exists in the code, so a leave request or a pay figure has no path into your Kastel at all.

The second barrier sits at the moment of writing. The connector names the fields it writes and writes those only. The record Factorial returns is never kept, and a field arriving in an unexpected shape is omitted rather than copied across as it stands. An unforeseen record produces a shorter entry.

The connector writes nothing into Factorial. The code holds one read verb, no write method appears in it, and the base address is pinned to a dated version of the interface so that a change at the publisher never passes without review. The same caution shapes the PayFit connector, with a different mechanism, because payroll and HR systems do not expose the same objects.

An org chart that prefers a gap to a guess

Factorial does not write the manager's name in an employee's record. It writes the id of another employee. The connector therefore rebuilds the reporting line itself, enumerating the whole directory on every pass to match each id to the name it belongs to.

When a manager id matches nobody in the enumerated directory, the manager line is omitted. No raw id is written in its place. The same rule holds for the department, which Factorial expresses as a link to a team. If that link does not resolve in the team list read on the same pass, the department line is absent, and the team is still read on its own as a full record.

The consequence is worth knowing before you connect. The org chart your AI reads can be less complete than the one Factorial shows you. It holds nothing that has not been resolved, which beats a guessed team link that nobody would know was wrong.

The questions HR leads ask us about this connector

Will an AI be able to see my employees' leave and absences?

No. Leave, absences and sick notes live in parts of Factorial that the connector never calls, and it calls the employee list and the team list only. There is no setting to watch on this point, since the address does not exist in the code.

What happens to the record of an employee who has left?

Factorial keeps a departed employee in its list, with a termination date. The connector does not read that date, so the record stays in your Kastel with their name, job title and manager, and nothing in it marks the departure. If the record does vanish from Factorial, it is then marked as gone at the source and kept until an administrator decides to erase it. A departure propagates through an administrator's decision, never on its own.

Why is the department missing on some employees?

Because their team link did not resolve in the team list read on the same pass. The connector leaves the line empty rather than guessing a department. Fixing the link in Factorial is enough, and the record updates on the following pass.

Who can read that directory inside my company?

The department your administrator has named, and nobody else. Factorial sets no read permission at the level of an employee, the key sees the whole directory, so the connector refuses to spread it department by department and never files a record in the employee's own department. With no explicit naming, the directory is filed in no department and waits for a decision. What Kastel protects, and how.

Can an AI change a record in Factorial?

No, no write path exists. The engine's Factorial reader exposes a read operation only, and the key it asks for serves nothing else. The question is closed at the level of the code rather than at the level of a setting.

What this connector does not do

It manages nothing. It books no leave, approves no absence, prepares no payroll, and never writes into Factorial. It gives an AI the structure of your company, and stops there.

The org chart can be incomplete. A manager whose id is not found in the directory read is omitted, a team link that does not resolve leaves the department line empty, and the field names of Factorial's interface are not all confirmed by real access on our side. The connector is written to produce a shorter record rather than an invented one, so what your AI reads can be less complete than what you see in Factorial.

An employee's departure is not visible. Factorial keeps them in its list with a termination date, and that date is not read. Their record therefore goes on existing in your Kastel, with their name, job title and manager, like the record of someone still in post. An administrator decides to erase it, and the same holds for a record that genuinely vanished from Factorial, which is marked as gone at the source and kept in the meantime.

A name tied to a job title, a team and a manager remains personal data, with no pay figure involved at all. And the directory is read as one block. Since Factorial sets no permission at the level of an employee, there is no way to open part of that directory to one department and the rest to another.

Tell us what your HR leadership is willing to make readable.

An employee directory is the kind of subject where a conversation beats a web page. We will tell you which fields come in, which are never requested, and what an administrator has to decide before the first pass.

Get in touch
Other connectors in detail
PayFit
Name, job title, department, manager. Never pay or the contract.
Lucca
Name, job title, department, manager. No salary, absence or contact detail.
Deel
Name, job title, department and manager. Never anyone's pay.

See the full connector catalogue