Victor lets a small program read people and groups from the company directory (AD) without opening the firewall, and keeps offices and cost centres up to date with a short script.
Administrator / integration engineer
Victor is the administrator at a company with 600 employees. Every login and group lives in the company directory on the company’s own servers (Active Directory, or AD). New hires appear there on Monday morning, and in every other system whenever someone remembers. Elena, who is responsible for security, will not open any door in the firewall from the outside (an inbound port). Victor wants people and groups in AverDesk without keeping a second list by hand.
HR created four accounts on Friday, one of them for a new warehouse worker. Victor has a note: add them, put them in the right teams, link their laptops. He has had this note before. Every time, someone ended up with access they should not have had, or without access they needed.
Victor installs a small program inside the company network, on Windows or Linux (an agent). It only connects outwards to AverDesk; nothing connects in, and no door in the firewall is opened. Elena agrees. The agent reads the directory without changing anything (AD or LDAP, the standard way to talk to directories, in read-only mode): people and groups, nothing else. It runs under a domain service account whose password is not stored anywhere.
Each directory group becomes a team in AverDesk. Teams decide which equipment people can see, and roles decide what they may do with it. The new warehouse worker is in the “Warehouse” group, so she sees the warehouse scanners and nothing from accounting. When someone leaves the group, the access goes with them.
Offices and cost centres are kept in the HR system, not in the directory. Victor writes a short script in n8n (an automation tool) or plain Python. It sends the data to AverDesk through its programming interface (the OpenAPI 3.1 API, with a header that names the organization). Records are created or updated in one batch, and repeating the same run changes nothing (idempotent bulk upserts). So running the script twice creates no duplicates, and a failed run can simply be started again.
Each sync leaves a trace in the history of every field. A person counts once, even if they work in two organizations. Victor’s Monday note is gone: the four new hires were in AverDesk, in the right teams, before he finished his coffee.
