glad.Add to Slack

Product

AI agent audit log and governance for teams, on every plan

Glad gives admins an AI agent audit log on every plan, the free trial included. It records who signed in, who changed a role, which apps were connected, what Glad ran, and who approved it. Filter by person, kind of action and date, and export it as CSV. Entries cannot be edited and are kept for 400 days. Admins also set per-person task limits.

Governance for an AI teammate comes down to three questions: who may do what, what happened, and how much it can use. Glad answers each one in the console, without an enterprise contract.

Add to SlackSee pricing

Last updated 9 October 2026

Example

Audit log
  • Oct 9, 14:02Glad

    Asked for approval to run a tool

    Gmail, send email

  • Oct 9, 14:03Sam

    Approved a request

    Gmail, send email

  • Oct 9, 14:03Glad

    Ran a tool

    Gmail, send email

  • Oct 9, 16:40Priya

    Changed a member's task limit

    From 40 to 60 tasks

Example only: four audit log lines as they could look for one afternoon.

How it works

  1. 1

    Roles come from Slack.

    The person who installs Glad becomes an admin. Anyone in your Slack who talks to Glad shows up as a member. Everyone signs in to the console with Slack; there are no passwords.
  2. 2

    Admins decide what Glad can touch.

    They turn on the apps and tools Glad may use. Each person connects their own accounts, and sharing a connection with the team needs an admin's approval.
  3. 3

    Admins set limits.

    A default monthly task limit for every member, and a different limit for any one person.
  4. 4

    Glad works, and the log fills.

    Every tool Glad runs, every approval, connection, role change and sign in becomes one plain line with the time and who acted: a person, Glad, or the system.
  5. 5

    Admins review.

    Filter the log, export it, and get a Slack message when the workspace reaches 80% and 100% of the month's tasks.

What you can do

The audit log (admins)

  • See every security and money action, newest first, grouped as Sign ins, Roles, Connections, Approvals, Tools, Memory, Billing and Data.
  • Filter by person, by kind of action, and by time: the last 24 hours, 7 days, 30 days or 400 days.
  • Export as CSV with the current filter. The export itself is logged too.
  • Trust what it says. The console states it plainly: "Nothing is edited or removed. Kept for 400 days."
  • Keep secrets out. The log never holds a secret or the text of a memory. It records that a memory changed, never what it said.
  • Keep history after people leave. Their lines stay, shown as "A former member".

Per-person limits (admins)

  • Set a default limit per member, counted in tasks per billing period. Leave it blank for no limit.
  • Override it for one person, and see what each person has used. An "At limit" badge shows who has reached theirs.
  • Know what happens at the limit. For that person, Glad still talks but uses no tools until the billing period resets. Everyone else keeps working.
  • See every change. Each limit change is logged with the old and the new value.

The workspace cap

  • Get warned early. Every admin gets a Slack message at 80% of the month's tasks, and again at 100%.
  • No surprise bills. At 100%, Glad still talks in Slack but uses no tools until the next month, an upgrade or a top-up. There is never an overage bill.
  • Everyone can see usage. The Spend page shows the plan and usage to every member. Only admins upgrade or buy a top-up.

Roles and data

  • Change a role between Admin and Member in Settings. The last active admin cannot be removed.
  • Approve or decline shared connections, and revoke a connection.
  • Export all workspace data as one file, or delete the workspace.
A task is one request where Glad uses a connected tool or makes a document. Talking, approving and runs that deliver nothing are free. A /deep run counts as 5 tasks.

Who controls what

Admins have the audit log and its export; Settings (members and roles, limits, plan and billing, approval rules, data export and workspace deletion); which apps and tools Glad may use and which connections are shared; shared memories; and approvals on shared connections.

Members use Glad in Slack, connect their own accounts and ask to share one with the team. They see their own memories and every shared memory, their own approvals, the Spend page and the Apps page.

Members do not see the audit log.

How other tools handle it

All facts below come from each vendor's own documentation, checked on 2026-10-09.
GladAudit logIn the console, admins onlyPlanEvery planExportCSV from the consoleKept for400 days
ViktorAudit logAudit Logs tab, admins onlyPlanEnterpriseExportPublic API only, with daily integrity digestsKept for400 days
LindyAudit logAudit logsPlanEnterpriseExportNot stated in the pages readKept forNot stated
Slack (including Slackbot actions)Audit logAudit logs, for Org Owners and Audit Logs AdminsPlanEnterpriseExportCSV and an APIKept forUp to two years
Claude TagAudit logActivity page for Owners; "no per-action log of every task and who asked"PlanTeam and Enterprise (beta)ExportNetwork events exportKept forNot stated
Where others are better today: Viktor offers its log through an API, with daily integrity digests "so you can check that no event was added, removed or edited", and an approval history that records the policy in force. Slack keeps audit logs for up to two years, has an API and anomaly detection, and marks actions Slackbot took for a user. Lindy has SOC 2 Type II and HIPAA, and adds SSO on Enterprise; Glad has no certificate yet.

Why it matters

NIST's AI Risk Management Framework calls Govern "a cross-cutting function that is infused throughout AI risk management". Govern 2.1 asks that "Roles and responsibilities and lines of communication related to mapping, measuring, and managing AI risks are documented and are clear to individuals and teams throughout the organization." Govern 1.5 asks for "Ongoing monitoring and periodic review".

The EU AI Act sets a similar floor for high-risk AI systems: they "shall technically allow for the automatic recording of events (logs) over the lifetime of the system" (Article 12).

This page does not claim compliance with either, and does not say Glad is a high-risk system. These rules point the same way: clear roles, a record of what happened, and someone who reviews it. Glad gives admins the roles, the log and the limits to do that on any plan.

Limits

  • No audit log API and no export to a SIEM. The export is a CSV download from the console.
  • No outside proof that the log was not changed, such as signed digests.
  • Retention is fixed at 400 days.
  • Search matches the person's name only, not the tool or the text.
  • No log of what Glad read, such as messages or files, or of who saw an answer.
  • Admins only. Members cannot see their own activity in the audit log; they see their own approvals on the Approvals page.
  • No SSO and no SCIM. Sign in is with Slack only.
  • No SOC 2 report yet. It is planned. Glad has no other certificate either.
  • No setting to make a tool ask first that would otherwise run.
  • Slack only today. Microsoft Teams is planned.

Questions

What is an AI agent audit log?

It is a record of what an AI agent did and who decided, kept so someone can check it later. In Glad it shows each tool Glad ran, each approval and its decision, and the admin changes around them, with the time and who acted.

What does Glad's audit log record?

Sign ins, role changes, connections, approvals, tool runs, memory changes, billing events such as limit changes and the 80% and 100% notices, and data exports or deletion. It never records secrets or the text of a memory.

Is the audit log only on an enterprise plan?

No. It is on every plan, including the free trial (50 tasks) and Starter at $49 a month. Every feature is on every plan.

Who can see the audit log, and can it be edited?

Only admins. Nothing in the log can be edited or removed, and entries are kept for 400 days.

How do I set AI usage limits per user?

In Settings, Limits, set a default task limit for every member, then override it for anyone who needs more or less. At their limit, Glad still talks with that person but uses no tools until the billing period resets.

Is Glad SOC 2 certified?

No. No SOC 2 report yet. It is planned. We never train on your data, your data is stored in the EU, and Glad is GDPR-ready: a DPA, a public subprocessor list, SCCs, and export and delete.

Keep reading

See it in a real task

Try Glad in your Slack.

50 free tasks. No card needed.

Add to Slack