Every agent run, open to audit
Agent History keeps a record of everything the room’s AI project manager was asked to do: who asked, each step it took, what it changed in the room and the agent credits it used. Open any run to review it, share it or carry the conversation on.
2 turns in this conversation
- Listed sprint tasks4:12
3 not done in “Northwind launch”
- Moved 3 tasks to “Northwind · wk 2”4:12
Dev K ×2 · Arjun M ×1
- Posted in room chat4:13
“3 tasks moved to next sprint” · @Dev @Arjun
An agent you can check up on
An autonomous agent is only as useful as your ability to see what it did. Every run in the room is kept as a timeline, not a summary.
- 01
Who asked, and when
Each run shows the person who gave the command, or an Automation tag when it was fired by one of the room’s automations, with the time it started.
- 02
Every step it took
Expand a run to see the plan as it executed: each step and each change, timed, with its status and outcome.
- 03
What it cost
Every run carries the agent credits it consumed, and one click opens your credit balance.
The full record of the room’s agent
Runs are grouped by day, newest first, with follow-ups kept together in one conversation.
Grouped by day
Runs are listed newest first under day dividers, so you can scroll back to what the agent did on any date.
Step-by-step timeline
Each run opens into its timeline of steps and changes, with times, statuses and outcomes, and the agent’s response.
People and automations
Runs started by a person show their name; runs fired by an automation say so, and open the automation that fired them.
Reply and continue
Follow-ups are stitched into one thread. Reply on a run for 24 hours after it ran and the agent picks up where it left off.
Credits on every run
Each run and each conversation shows the agent credits it consumed, next to the time it ran.
Retry, copy, share
Retry a command that has just failed, copy or share an answer, or have it read aloud. Runs tied to a task open that task.
Every command you gave, in one place
Inside a room, Agent History shows that room’s runs. The Commands tab gathers your own commands from every room into one feed.
A chip for each room
Each row says which room the command ran in, and opens that run beside the list.
Search your commands
Filter the feed by what you asked or by the room’s name.
Ask from the same screen
The prompt bar sits above the feed, so a new command is one line away from the history.
Private stays private
A run you keep private is visible only to you. You can switch your own runs between private and public after they run.
Commands · all rooms
- Northwind launchToday · 4:12 PM−3
Move the Northwind tasks that slipped into next sprint
- HiringToday · 11:30 AM−2
Schedule second-round interviews for the three shortlisted designers
- Private roomToday · 8:05 AM−1
What did I commit to in this week’s meetings?
- FinanceYesterday · 6:20 PM−2
Add a Due date field to the Invoices table and fill it from the notes
- Northwind launchYesterday · 3:02 PM−2
Write a launch brief from the kickoff minutes
From a command to a reviewed record
Agent History is the end of every run and the start of the next one.
A command is given
Someone asks the agent to plan and execute, or an automation fires.
The run is recorded
Every step, change and credit is kept on the run as it happens.
The output is checked
Open the task a run worked on, or the items it created, straight from the run.
Patterns become automations
A command you keep repeating can be built into an automation for the room.
Who runs on Agent History
Know what the agent changed
Before the Monday review, scroll the week’s runs to see which tasks, meetings and notes the agent created and for whom.
Keep an eye on automations
Runs fired by automations sit in the same history as people’s commands, so a misfiring rule is easy to spot and open.
See where credits go
Every run shows its credit cost, so it is clear which commands and conversations consume the room’s agent credits.
One room, every app reading the same state
Agent History, answered
The command as it was asked, who asked it or which automation fired it, the time, each step and change the agent made with its status and outcome, the agent’s response, and the agent credits it used.
Public runs are visible to the room. A private run is visible only to the person who gave the command, and they can switch it between private and public afterwards.
Yes, for 24 hours after it ran. Reply on the run and the follow-up streams into the same thread.
The error is shown as a step in the run’s timeline. A command that has just failed offers Retry, and you can reply to the run to tell the agent how to try again.
Yes. The Commands tab is a feed of your own commands across all your rooms, each tagged with the room it ran in.
Deploy an AI project manager with Agent History
Every agentic room comes with its apps and its AI project manager. Start free, or walk through it with us.