An AI unused license detection agent pulls the seat roster for an application and lines it up against the activity records available for it, such as last login or last action dates.
Reads assigned seats for one application from the admin export.
Joins each seat to its last recorded sign-in or action.
Compares the idle gap with the window set for that application.
Moves shared and leave-of-absence seats to a review-only list.
The software owner confirms which seats are safe to reclaim.
Records reclaimed and kept seats against the renewal date.
It applies the inactivity window and the exceptions your team has defined, for example leave of absence, shared service accounts or seats reserved for a new hire. Seats that match an exception are listed separately instead of being flagged.
It then writes a short report per application for the owner, with the seats, the evidence behind each flag and the renewal date, so the decision can be made before the next billing cycle.
Seat counts are billed from the vendor's roster, while activity lives in a different system, so nobody sees both side by side without a manual export.
Assignments sit in the admin console and activity sits in logs or identity records, so matching them means joining by email and handling mismatches.
A login, an API call and an opened document all count differently, and each vendor reports a different one.
Service accounts, seasonal staff and people on leave appear idle but should not always lose access.
A finding that arrives after the renewal notice window has closed can no longer change what you pay.
Read the current seat list for the application from the records provided, including assignee, role and assignment date.
Join each seat to the last-activity date for the same person, normalising email aliases and merged accounts.
Compare the gap since last activity with the inactivity window your policy sets for that application.
Move seats that fit a defined exception, such as a shared account, to a review-only list with the reason stated.
Produce a per-application list with evidence and the nearest renewal or notice date.
Bring a small sample and watch Flo work through it.
| Aspect | By hand | With the agent |
|---|---|---|
| Evidence | Exports from two consoles are joined by hand in a spreadsheet. | Roster and activity are joined automatically, with the source of each flag kept. |
| Exceptions | Service accounts get flagged by mistake or missed entirely. | Defined exceptions are sorted out first and shown with their stated reason. |
| Timing | Review happens when someone remembers, often after the notice period. | Lists arrive ahead of each renewal with the notice date shown. |
| Consistency | Each owner picks their own idle cut-off. | One window per application is applied the same way every cycle. |
| Follow-through | Reclaimed seats are rarely recorded anywhere. | Each decision is logged so the next cycle starts from known facts. |
| If | Then | Who decides |
|---|---|---|
| A seat shows no activity beyond the window | List it as a reclaim candidate with its last activity date | Application owner decides |
| The assignee is on recorded leave | Move the seat to review-only with the leave noted | Line manager confirms |
| The account is marked shared or a service account | Exclude from flags and list it with the stated reason | Application owner |
| The vendor reports no activity data for a plan tier | Mark activity as unknown instead of idle | Application owner is told |
| The seat was assigned very recently | Hold it back from the list until the window has passed | No action needed |
| A renewal notice date is close | Raise the application to the top of the queue | Owner and finance contact |
A person signs in with an alias or an old address, so the seat looks idle. The agent flags the mismatch for the owner instead of guessing which account is theirs.
Some staff only use a tool at quarter end or during audits. The owner can tag them so the agent lists them for review, never as plain reclaim candidates.
Reclaiming one seat may not lower the price if the contract sets a minimum. The agent shows the minimum next to the finding and the buyer decides.
Administrators may hold seats they rarely touch but must keep. The agent lists them separately and the owner confirms.
Talk it through with the team. We will map it to your steps and show where a person stays in control.
Pick a single tool with many seats and clear activity records, and run it in report-only mode.
Owners review every flag against what they know and note where the agent was wrong.
Write the exceptions and windows that came out of the review, then add more applications.
Schedule lists ahead of renewal notices and keep a person approving every reclaim.
Read this per application, not as a company-wide average. A high share on one tool points to a renewal conversation, while a low share across the board suggests assignments are in good order.
Quick answers to what people ask most about unused license detection.
Book a demoIt is the process of identifying software seats that are paid for and assigned but show no meaningful activity. The aim is to give owners evidence before a renewal so they can reclaim or reduce seats.
There is no single answer. Many teams choose a window tied to how the tool is normally used, with longer windows for quarterly or seasonal tools. The window is a policy your team sets.
No. It prepares the list and the evidence. An owner decides and an admin changes access in the application.
It needs the seat roster and some form of activity record per user, plus the renewal date and contract terms. Where activity is missing for an application, the agent says so instead of guessing.
License optimization is the adjustment of licence types, tiers, counts and plans so they match how software is actually used.
→Seat rightsizing is adjusting the number and tier of software seats so they match how people actually work.
→License utilization is the comparison between the licences or seats an organisation pays for and those people actually use.
→A SaaS security review is a check of whether a software vendor's security practices and data handling meet your requirements before or after purchase.
→Bring a small sample of real items. We will show what Flo handles on its own and where it hands over to your team.
