Skip to main content

Record Guide

This guide covers what to look at once it has run: how to read the results, how to trace what happened, and how to revisit things from the track record.

How the records work is covered in Records and metrics.

Where a section has a place on screen, you will find where to set or check what it asks you to decide at the end of it.

Where to look​

There are five places.

  • Overview — from the top, “Project overview” (a description written by AI from the purpose, steps and Agent settings, or the project description; not shown when there is neither), “Results of this work” for the period you choose (a sentence on the results and the metrics), and “Workflow shape”, the flow of steps and agents (its right-hand column switches between “Basic info”, “Metrics”, the figures for each step over the same period, and “Latest artifacts”, which is not shown to people who open the project only through the organization operations role); for a cycle, the band of the last 24 hours for each agent comes below it, and at the bottom is “Project info” (a one-line summary; “Show All” opens all of it). Pressing an agent in the band takes you to the execution history narrowed to that agent
  • Execution history — the runs. Ones still going appear here too. The “Trigger” column shows what started each one (Manual, Scheduled, External API, Email, Sweep, Webhook) and, where a person or an app started it, their name (an external sender has “External ([EXT])” in front). Review requests that have been decided appear as rows too, with “Reviewed by (name)” or “Decided automatically” in the agent column. The result column also shows the first paragraph of the agent's output and, where there are artifacts, how many (such as “Artifacts: 3”). A row for an external automation shows that it was sent, with the HTTP status; a row for a conditional branch shows the lane that was chosen (or joining directly), or that a review was opened for a person to choose. You open one from here. From the run record of a workflow step, “View this run in history” narrows the list to the records of that workflow run. You can also narrow it by “What to review” (instructions, Skill, handover from the previous step, connector connection, execution rules and so on). A run started by rerunning a sweep is marked “Rerun”
  • Workflow history — the workflow runs. It appears for workflows only (a cycle has no run that spans the steps). It is not narrowed to what is still going. Each agent placed on a step has one dot (a review step or a step with no agent has one dot per step); redoing does not add dots, and the latest state is shown: “Planned” for steps not reached yet, “Not run” for steps that did not run because the run failed or was rejected, and “Cancelled” from the step that was stopped onwards when the run was cancelled (hover over a dot to see it as “step name (agent name): state”, or “step name: state” for a review step or a step with no agent). A run in progress with a deadline shows “Deadline: in ◯”, and “Past deadline” once it has passed (the run does not stop; see the section below). A run closed by a rejection or a failure shows “Resume” to owners, operators and people with the organization operations role (not while an admin has stopped the project; Review Guide)
  • Action needed — what needs a person. Review requests not yet decided, runs that did not end well, and runs past their deadline or not moving, in order of urgency (next section)
  • KPI — the figures over a period. It sits under “Results of this work” in Overview

In the project list, the cards under “My Projects” also show “Completion rate (7d)” for the last 7 days and the number running (a stopped project shows “Stopped” and its status). The rate is counted the same way as the agent completion rate. Cards under “Public Projects” do not show these.

Review requests across projects are seen under “Inbox” on the left > “Review requests” (Review Guide). Switch to “Action needed” and you see, from the “Action needed” rows of the projects you take part in, only the rows assigned to you.

Figures across projects are seen under “Insights”. It shows the figures for the projects you are a member of taken together, and a list by project (how they are counted is under “Reading the figures” below).

Decide who checks runs that did not end well and insufficient records, review requests left undecided, and the authentication status of connectors, and how often (what does reach you as a notification is in the Connection Guide). For the applications workflow, the owner might look at “Action needed” each morning and nudge the reviewers about what is left in review requests. Insufficient records and connector authentication do not appear in “Action needed”, so check them in the execution history and on the connectors screen.

Where on screen Opening a project gives you the Overview / Execution history / Workflow history / Action needed / Library / Settings tabs (Workflow history for a workflow only). KPI is under “Results of this work” inside Overview, and the record of operations inside the project is under “Records (audit trail)” inside Execution history. Figures across projects are under “Insights” at the top of the screen before you open a project

Clearing what needs a person (Action needed)​

“Action needed” lists what needs a person, one row each. There is nothing to set up. Each time, it is built from the review requests, runs and sweeps as they stand at that moment.

KindWhat is listedAssigned to
Awaiting approvalReview requests not yet decidedReviewers
OverdueReview requests past their deadline and not yet decidedReviewers
FailedWorkflow runs in progress where the agent on the current step failed and no review request is openOwners
Awaiting reworkRuns sent back where the step they went back to has not run yetThe reviewers of the review request that sent it back
Sweep failedIn a cycle, placements whose latest sweep ended “Failed (reported by agent)”, “Failed”, “Not allowed by rules” or “Timed out” (asking for a rerun takes it off; sweeps stopped for a connection confirmation are not listed)Owners
Past deadline (run)Workflow runs past their deadline and not finishedOwners
StalledWorkflow runs that have not moved beyond the “Hours before a run counts as stalled”Owners

The rows are in order of priority (P1 to P4). Priority is set mechanically and cannot be changed by hand. Within the same priority, the nearer deadline and the older row come first.

  • P1 — Overdue, Past deadline (run)
  • P2 — Failed, Sweep failed, Stalled, and Awaiting approval with less than 24 hours to the review deadline
  • P3 — Other Awaiting approval, Awaiting rework
  • P4 — Awaiting approval with no deadline

The project's “Action needed” tab lists every row, whoever it is assigned to. Members and people with the organization operations role see every row too. “Action needed” in “Inbox” on the left has only the rows assigned to you. Both can be narrowed by the kinds above. Up to 100 rows are listed at a time; beyond that, “◯ more not shown” appears.

Opening a row takes you to its screen. A review request row opens the review request; a workflow run row (Failed, Awaiting rework, Past deadline, Stalled) opens that run's screen (“Case”); a failed sweep opens its run record.

Acting on several rows at once​

Select rows and “Bulk actions” appears. There are six actions. Each runs one by one on just the selected rows it applies to (the number of rows it does not apply to is shown).

ActionRows it applies toWho can
ApproveReview requests awaiting approval or overdueReviewers and their delegates
Reject, Send backReview requests (a connection confirmation can only be rejected). Enter the reason once and the same reason goes on every one. Each is sent back to its default stepReviewers and their delegates
Extend deadlineReview requests (not connection confirmations). Enter the new deadline (a date and time later than now) (Review Guide)Reviewers, their delegates, owners, operators and people with the organization operations role
Rerun sweepSweep failedOwners, operators and people with the organization operations role
Reassign approversReview requests of a workflow run. Choose the new approver from the members of that project; rows from more than one project cannot be reassigned together (Review Guide)Owners, operators and people with the organization operations role

The result is shown for each row. Besides “Done”, there are “Already decided” (someone else decided first), “No permission”, “Conflict (content was updated)”, “Decide individually”, “Not found”, “Failed”, “Stopped by admin” (an admin has stopped the project) and “This action isn't available to the operations role”. For approving, rejecting and sending back, even an owner gets “No permission” on a review request where they are neither a reviewer nor a delegate. Extending the deadline goes through for owners, operators and people with the organization operations role, even when they are not reviewers. A review request that cannot be decided in bulk, such as one with destination options, is left out of the bulk decision and a note says it cannot be decided in bulk; open it and decide it on its own.

When a periodic sweep of a cycle ends “Failed” or “Timed out”, the platform retries it automatically, once. A sweep that failed after writing to an external system is retried too. The retry runs as a new execution shortly after the sweep ended (its attempt in the run history is 2). A sweep that ended a while ago is not retried. Sweeps that ended “Failed (reported by agent)”, “Not allowed by rules” or “Failed (stopped)”, sweeps stopped for a connection confirmation, sweeps started by a webhook or an email, and retried sweeps are not retried. If the retry also fails, decide on a rerun from the “Sweep failed” row in “Action needed”.

Rerunning a sweep runs the placement once with its current settings. If the original sweep ran from a webhook or an email, the same body is attached again. A sweep can be rerun only once. The run it starts is marked “Rerun”, and its run record shows which run it reran. The Records (audit trail) keep it as “Sweep rerun requested”.

Where on screen The project's “Action needed” tab. Across projects, “Inbox” on the left > “Action needed”. The bulk actions are under “Bulk actions”, which appears when you select rows

Deadlines and stalls (workflow)​

A workflow run can be given a deadline. Even past its deadline, the run does not stop. It is listed in “Action needed” under “Past deadline (run)”, and the owners are notified.

  • Default deadline — choose “Default run deadline (hours after start)” on the Settings tab and every run started after that gets a deadline that many hours after it starts. The default is “None”
  • Per-run deadline — owners, operators and people with the organization operations role can set, change or remove the deadline on the screen of a run in progress (opened from a row in “Action needed”)

A run that is not moving is reported as “Stalled”. That is a run that has not moved for the “Hours before a run counts as stalled” since its last step finished (or, if no step has finished yet, since it started). The default is 24 hours. A run waiting for a review, and one whose cancellation has been asked for, are not counted. A run past its deadline is listed under the deadline instead.

A notification goes out once per deadline and once per stall. Who receives it and how to turn it off are in the Connection Guide.

Where on screen The default deadline and the stall hours are under the Settings tab > “Execution rules” > “Execution” > “Default run deadline (hours after start)”, “Hours before a run counts as stalled” (only an owner of a workflow can change them). A per-run deadline is set with “Set a deadline”, “Change the deadline” and “Remove the deadline” on the run screen opened from “Action needed”

Reading a single run​

Pick a time in the execution history and that one run's record opens. This is the run record.

At the top of the “Record” tab, the run is read out in a few short sentences. In order: what started it (and that it is a rerun, if it reran a sweep, or that it started again after a connection confirmation was approved), the calls to connectors, the review requests, the result and what to review. The sentences are built only from the facts in the record; they do not read the body of the exchange or the contents of artifacts.

Below that comes “What was produced”. It holds the run's agent output (for a workflow step, the first 2,000 characters) and artifacts (it does not appear for a run with neither). On a run that went on after the output was corrected in a review, the corrected text comes first and the agent's output is headed “Original output”. Pressing an artifact's name opens a preview of the version as of the record (or of the current version, when a later run overwrote it and that version is no longer kept), where it can also be downloaded. Each artifact has the following.

  • “Register to the project library” — puts the artifact in the project library, in its version as of the record (members can do this too, if they may use “File attachment” under the feature permissions the administrator sets). When a file of the same name is already in the library, it becomes “Add as a new version of "(file name)"” and goes in as a new version of that file. What was put in from this record shows “Registered”, and pressing it opens that file in the library. An artifact whose version as of the record is no longer kept shows “Overwritten by a later run” and cannot be put in
  • “Save to the library under this name every run” — shown to owners only. It puts the artifact in the library and adds it, under the same name, to “Files synced with the library” of the placement on this run's step. From the next run, what the agent writes under that name is kept as a new version in the library. In a tenant where changes to definitions need approval, this starts only once the change is approved (Design Guide). It is shown for files directly in the workspace that are not synced yet. An artifact whose version as of the record is no longer kept goes in as its current version (without the “Registered” mark)

When a step has “Values and file roles passed to the next step”, the values and files returned in that run, and those not returned or not found, appear below “What was produced” (a value cut at 1,000 characters shows “(cut at 1,000 characters)”). They also appear on the step rows of a workflow run.

“Chat with this result”, next to the “What was produced” heading, lets you follow up on the result. It opens a new chat with the agent's output (for a workflow step, up to 20,000 characters) in the message box and the artifacts attached (nothing is sent yet). The output goes in wrapped with a note to treat it as data, not as instructions. The artifacts are copied to the library of the person who pressed it. The full conversation is not passed on. Members can do this too. Where some artifacts could not be attached, their number is shown.

Where a run did not end well, a “Failure summary” appears, below “What was produced” when that is shown (not for a run stopped for a connection confirmation). It gives Cause, Impact and Next action, and What to review where one was recorded — and the next action points at where to look (connector settings, execution rules, step settings and inputs, or what was done). Read this first.

“What to review” is where the fix most likely lies. It shows as “Instructions (prompt)”, “Settings and connections (harness)”, “Restricted by an administrator or the system”, “External service” and the like, and when the agent's report has no comment, or the failure came from the platform, an explanation of what to review comes with it. The next-action buttons follow What to review where it narrows things down, and otherwise follow the failure category.

The record is laid out as five questions.

  • Request — which step it ran as the agent for, and what started it. Who set it up (or, where an app started it, which app; for a workflow started by email, who sent it), when it was accepted, and the attempt number from the second attempt on. Where there are input files set on the agent or files received from the previous step, they are listed separately under “Input files set on the Agent” and “Files received from the previous step”. On a step set to “Run at the same time”, “Agents run at the same time” lists how each agent of the same round is doing; on a workflow run with lanes, once the split has begun, every record of that run shows how each lane stands now (waiting for the other lanes, joined and so on) — an earlier step's record shows the current state too
  • Evidence — how many reads, how many kinds of tool were used, and how many external-service calls there were, of which how many failed (counts per connector are shown too). Whether the connectors connected shows as “Connectors connected: ◯/◯”, with the ones that did not connect and, separately, counts by reason
  • What was done — the number of tool calls, a list of which tool did what, and what became of each (performed, attempted only, denied, effect confirmed). Where the matched rule was recorded, a denial comes with the “Matched rule”. The agent's output and the files it produced appear under “What was produced”
  • Confirmation — split into “Questions from the Agent” and “Review requests”. In a project's runs the agent does not ask a person anything partway through, so Questions from the Agent shows “The agent asked no questions or confirmations of a person in this execution”. Review requests appear in the run record of the review step's row. A review request raised from the result of a “Run by Agent” step also appears in the record of that agent's row. A decided one shows the decision and who made it, and one not yet decided shows “Waiting for confirmation”. Connection confirmations appear in a separate “Connection confirmation” part. On the stopped agent's row and in a sweep's run record, it lists the connector's name, the state (Expired, Voided and so on) and the purpose the agent declared; selecting the name opens the review request
  • Result — how it ended. When the run simply completed and there is nothing to add, such as a reason or an error, this item does not appear (the badge in the heading shows the result)

Denials can appear under “What was done”. Those are operations stopped by the execution rules or by limits fixed on the product side (Execution Rules Guide).

The “What was done” list opens under “Operation details”. Reads and external-service calls appear there too. But the substance of an operation — the values passed in, the body written — does not.

In a run record, pressing a name under “Input files set on the Agent”, “Files received from the previous step” or “Artifacts” opens a preview, where it can also be downloaded. This is for members of the project. The versions shown are the latest one and those written since the start of the first run that recorded the file as an artifact.

On a run that was handed files from the project library, “Request” lists them under “Files handed over from the library”, with their versions. Where the agent rewrote such a file, “What was done” shows under “Saved to the library” what became of it (New version saved, Unchanged (no new version), Failed to save, Not registered (kept as an execution output)). How the library works is in the Design Guide.

The “Run info” tab has “Settings at run time”. These are the values saved at run time (such as the basic rules at run time and the model), and changing the settings later does not change this view. They are checked against the values the execution platform saved for the same run, and an item that differs shows “Differs”, with a notice at the top of the run record as well. An item with no value shows “Not recorded”.

Unless the agent has been removed with “Remove from project”, “View current settings” in the heading shows that agent's current settings (removing it only from a step does not hide it). Its goal (instructions), guardrail, and the names of the playbook, skills and input files set on it are shown read-only. These are the settings as they are now, not as they were when it ran. “Edit Agent” is shown only to the agent's owner.

The review requests under “Confirmation” show only the decision and who made it. For the content, press the name to open the review request (Review Guide). The body of the exchange is not shown here either.

The “Timeline” tab shows “Execution flow” (the events from start to end and the agent's operations, in order of time) and the “Execution platform record summary”. The “Run info” tab shows “Settings at run time”, “Execution platform records” (the values saved by the execution platform, shown as they are; items without a record show “Not recorded”) and “Identifiers”.

Some runs have a gap in the record. Where it says no result was recorded, it is counted as neither success nor failure.

Where no value was obtained, a reason is shown. How blank, zero and “Nothing to do” differ is in the records and metrics concepts.

ShownMeaning
Not recordedNo record of that item remains. It is left blank rather than filled from the current settings
Result not recordedThe result of a finished run was not recorded. It is counted as neither success nor failure
Insufficient recordsSome runs are missing their result. The KPI is a provisional value that leaves those runs out (“Records incomplete”, “Insufficient records”)
Could not loadThe information could not be loaded. Wait a while and open it again
No permissionYou do not have a role that can see that information
No data, or you don't have permissionWhat you opened does not exist, or you cannot see it (the two are not told apart)
Nothing to doNot a gap in the record but a result. It means the run never started
Not measuredThat value is not something measured
Legacy recordA record kept in an earlier form. Some things, such as whether anything was written, are not known
“The recorded reason (result) cannot be interpreted on this screen”The recorded reason or result is not among the words this screen knows. For a result, it is counted as neither success nor failure
Deleted (cannot be viewed)It was deleted and cannot be viewed
Delivery unconfirmedIt has not been confirmed that a notice that was sent (such as a review request email) arrived
This information is not available yetThat information cannot be shown on screen yet

Where on screen The Execution history tab > a time in the list > the run record. The “Record” tab starts with the read-out, followed by “What was produced” (with “Chat with this result” next to its heading); where a run did not end well, a “Failure summary” follows; below them come “Request”, “Evidence”, “What was done”, “Confirmation” and, when there is something to add, “Result”. A run’s record is split into three tabs: “Record”, “Timeline” and “Run info”.

Do not misread the words for the result​

How it ended is given in seven words.

  • Completed — it means the agent neither reported that it could not finish nor said so in its final result
  • Failed (reported by agent) — the agent itself reported that it could not finish, or said so in its final result (after wrapping up at the run time limit, it is “Timed out”)
  • Failed — it ended abnormally
  • Failed (stopped) — a person or the platform stopped it
  • Timed out — it reached the time limit (including when it wrapped up at the run time limit and could not finish)
  • Not allowed by rules — the platform forbade the run itself
  • Nothing to do — there was nothing to act on, so the run never started

“Completed” is no guarantee that the work went well. The agent simply did not say that it could not finish, so check what was done under “What was done”. Since the substance of the operations is not shown, whether it was right as a piece of work is checked against the output itself.

There are three kinds of “Failed”. What you do about them differs, so read what is in the brackets: for “Failed (reported by agent)”, read the “Agent's closing report” in the “Failure summary” (permission, missing input, the connected service not responding, restriction by the execution rules, and so on); plain “Failed” ended abnormally, so read the reason given with it; “Failed (stopped)” was stopped by a person or the platform.

Tracing what happened​

Review decisions, changes to execution rules and members, and similar events are traced under “Records (audit trail)”. Each is listed with its time and who did it (the events that are recorded are in the Records reference).

They can be filtered by event type, and searched by word. They can be exported as CSV too, but only the event type filter carries into the CSV; the search term does not. The list shows 50 at a time, newest first; “Load more” reads earlier ones. When the CSV is cut off at the limit, the number exported and the total are shown.

Not everything was done by a person. “Agent”, “System” and “External sender ([EXT])” appear there too. An automatic action at a deadline is recorded as System (Review Guide).

Where on screen The Execution history tab > “Records (audit trail)”. “Search records”, the “Event” filter and “Export CSV” are in the same place

Opening the body leaves a reason behind​

Where a run has a body, you can open the exchange with the agent. Opening it requires a reason (up to 300 characters). Owners, operators and members alike can open it. “Open the full record” shows the body, read-only, on a separate screen. Some runs have no body, and those cannot be opened. The conversations created for step runs and sweeps are titled “[project name] agent name” (a run started by a webhook has “[webhook]” in front, and a sweep started by email has “[mail]”). These conversations do not appear in the chat list, counts or search of the project's owner at the time of the run. Opened in the chat screen, they show “This conversation can be viewed in the project's execution record.” instead of the body, and, if you are currently a member of the project, “View execution record” goes to the record. To follow up on a result, use “Chat with this result” under “What was produced”.

The reason and the viewer are recorded. “Export” on the run record works the same way and requires a reason. Who looked, when, and what for is kept.

“Export CSV” on “Records (audit trail)” also needs a reason (up to 300 characters). The export is kept, with its reason, in Records (audit trail) as “Execution records exported”. What it exports is the Records (audit trail) list only; it holds no bodies.

Who looked at a body is visible to the person concerned. They can check who has looked at the body of their own run under who viewed this record. The project's owners, operators and members, and people with the organization operations role, can check the same record.

A body holds the content of the work. Open it only where you need to.

Where on screen The run record > “Open the full record” / “Export”. The list of who looked is under “Who viewed this record” on the same screen

Reading the figures​

The figures for the results count one agent execution as the unit. You can choose the period (values are in the Settings reference).

🚨 The review figures count something different from the results. And how an automatic action at a deadline is treated differs by figure on top of that. Get this wrong and you misread how much people are doing.

  • Reviews (per workflow run) — counts only what a person decided
  • Approved by people and Send-back rate — what advanced automatically at a deadline affects neither figure. What was sent back or rejected automatically does
  • Send-backs, rejections, cancellations and Typical wait — these include automatic actions. “Decided on time” is the share of review requests with a deadline that the reviewer answered by the deadline; ones settled by an automatic action at the deadline are not counted

How each figure is counted is in the Results reference.

A retry is counted as a separate execution by the agent. The same agent running the same step twice makes two.

Some things are left out of the figures. Nothing to do, still running, and insufficient records are all excluded from the “Agent completion rate”. Runs that ended with no change (the agent reported that what it checked had not changed and did nothing; the result is “Completed”) are left out too, and they do not appear in the “Excluded” counts either. Runs stopped for a connection confirmation are left out too, and appear under “Excluded” as “Stopped for connection confirmation”. Where a run has no recorded result, the figure is provisional. Under “Results of this work” in Overview, “Records incomplete” appears in place of the figure; in “Insights”, a note says it is provisional (both give the number of runs with no recorded result).

The “Agent completion rate” is not the business outcome. It is the share of runs whose result came out as “Completed” (leaving out those that ended with no change). Runs that ended with no report at all count as completed, unless the final result said the agent could not finish, but timed out, stopped, not allowed by rules and failed do not.

In “Insights”, the agent completion rate, the counts and the typical duration are counted in the same way. They take the executions of the projects you are a member of together. The list by project also shows how many runs have insufficient records, and each row opens the execution history for the same period. The review figures do not appear there; see them in each project's KPI.

Where on screen The Overview tab > “Results of this work”. “Period” and “Export CSV” are in the same place. Figures across projects are under “Insights” at the top of the screen before you open a project, with “Period” in the same place

Revisiting it from the track record​

The clues for revisiting are the run records and the review track record. Choose what to fix from what you see.

What you seeWhat to fix
The “Failure summary” causeWhere to look first for each cause is in the Run Results Reference. If the execution rules stopped it, also see whether that operation should stay blocked, and whether the goal can avoid using it
Denials are listed under “What was done”Likewise the execution rules, or the goal (Execution Rules Guide)
Send-backs keep coming, with the same point in the reasonsThe agent's goal and skills
“Completed”, yet the deliverable cannot be used in the workThe goal's “to what state it finishes”, and the skills (Design Guide)
In a cycle, the same item is picked up again and againThe rules for the markers in the shared record (Design Guide)

How to try a fix before applying it is in the Design Guide.

From a workflow step run or a cycle sweep that did not go well, you can consult the AI for improvement ideas. The “Failure summary” shows “Ask for improvement ideas (alpha)” on runs that are “Failed (reported by agent)” with a recorded category, “Not allowed by rules” because a guardrail or an execution rule forbade it, “Timed out” on reaching the execution time limit, or closed because a connection confirmation was rejected or expired. On a workflow step you can also consult when the result is “Completed” but the reviewers rejected it, or when it has been sent back to the same step up to one short of the limit. What the AI reads is the closing report, the list of what was done (up to 80), the settings, and the body of the skills (the first five) and playbook you can read. It is also given the reasons for send-backs to that step and rejections, the results of connecting to connectors and where an operation was stopped, for a cycle the report of the previous sweep of the same placement, and the improvements applied to this placement so far (only the date and kind, the latest five) with the results since. A long body may be used only in part. The agent's output is not passed to the first consultation; if you choose to include it, it is passed from your next message (only in part when long).

The proposals come in three kinds — “Revise instructions” (the agent's goal and guardrail, a step's description), “Hand over to a person” (adding a review step, removing a placement) and “Add to Skill or Playbook” — and missing inputs or connections are shown as “Add missing input or connection”.

Members can consult too, but only owners can “Apply”. Applying saves the chosen changes on the spot, and they are used from the next run. In a tenant where changes to definitions need approval, runs keep using the previous version until the change is approved. That run's result stays as it was, and nothing is re-run automatically. Where the agent to change belongs to someone else or is also used in another project, the agent is copied and only this project's placement switches to the copy. If the same agent is placed on several steps in this project, a change to its goal or guardrail applies to every step where it is placed. A playbook owned by someone else is copied, and the agent is made to use the copy. A playbook you own is changed itself, even if it is used elsewhere. For a playbook you have published, in a tenant where publishing needs approval the change goes in as a publishing request. If a setting the proposal relied on changed after the consultation, that change is not saved and the changes after it are not applied (earlier ones have already been saved). Check the content and consult again.

An addition to a skill depends on who owns the skill.

  • A skill you own — it becomes a new version. It also takes effect on other agents that use the skill with “Follow the latest version”. If the tenant requires administrator approval for publishing and the skill is public, the new version goes in as a request and does not take effect, in this project either, until it is approved. Where the agent pins the skill's version, the new version is not used (to use it, choose the version again in the agent's settings)
  • A skill someone else owns — a copy is made for you with the addition, and the agent is switched to the copy. Where the agent is also used elsewhere, the agent is copied too and only this project's placement is switched

A skill containing code such as scripts cannot be added to. For a skill you own, it also cannot be applied when the version the agent uses differs from the skill's current version (check the skill's versions in the library).

Loosening the reins on reviews​

This section is about workflows. A cycle has no review steps, so this setting does not appear for one (Review Guide).

It is safer to start on "Stop and wait for a decision" together with "Notify and keep waiting", and loosen it once you have a track record. "Stop and wait for a decision" on its own is not enough: depending on what is set for after the deadline, it may advance, send back or reject automatically (Review Guide).

There are two figures to look at.

  • Approved by people — the higher it is, the greater the share of decisions a person approved
  • Send-back rate — the lower it is, the smaller the share of send-backs

Send-backs and rejections done automatically affect both of these. When what is set for after the deadline is "Send back automatically" or "Reject automatically", these two do not represent people's decisions alone.

Where the first is high and the second is low, you can change to "Notify and proceed after the deadline". Where send-backs keep happening, revisit the agent's goal before loosening anything (Design Guide).

The review method is changed per step, and only the owner can change it. It can be changed under Execution rules on the Settings tab or under “Check settings” for the review step in the editor, and both change the same setting; the KPI screen is there to show the track record. Who changed it and what they changed are recorded. A changed review method applies to runs started afterwards. Runs in progress keep the method they started with.

🚨 Loosening the review method does not allow an operation the execution rules block. The review method governs the flow of the work; which operations are allowed is a separate axis (Execution Rules Guide).

Where on screen The track record is under the Overview tab > “Workflow shape” > “Metrics” > the box of the review step. The change is under the Settings tab > “Execution rules” > “Step review method” > the change button on each row (in a workflow, for the owner only). “Open settings” in the box of the review step under “Metrics” also takes you to “Execution rules”. It can also be changed under Edit > select the review step in the diagram in the middle, then “Check settings” on the right > “Basic settings” > “Review method”

Confirming the effect​

The KPI period is chosen from set lengths counted back from today (Settings reference). It cannot be cut at the day you made a change. So export the KPI CSV before the change and keep it, then, once the same length of time has passed after the change, export it again and compare. What the CSV holds is in the Run Results Reference.

Alongside the Agent completion rate, check the deliverables against when the work is done, as settled for the work (Design Guide).

Distinctions worth keeping straight​

These are the distinctions that cut across the sections above.

  • Taking “Completed” for a success in the business sense. It means the agent neither reported that it could not finish nor said so in its final result. What was done is checked under “What was done”
  • Looking for the count of external-service calls under “What was done”. The counts of external-service calls and of their failures are under “Evidence”. The count under “What was done” is the number of all tool calls; the list itself also carries reads and external-service calls
  • Thinking the record follows an operation down to its substance. What you get is what was done — not the values passed in or the body written
  • Reading the review figures as one. They count different things, and they treat an automatic action at a deadline differently. “Approved by people” is unaffected by what advanced automatically at a deadline, yet it is affected by what was sent back or rejected automatically
  • Thinking that loosening a review loosens the operations. The review method and the execution rules are separate axes. A blocked operation stays blocked whatever the review method
  • Opening the body lightly. A reason is required, and who looked, when, and what for is kept
  • Looking for “Workflow history” in a cycle. It does not appear there. What is running is seen under Overview and the execution history
  • Taking a run with no record for a failure. It is counted as neither success nor failure. In the figures it is left out as insufficient records
  • Taking “Stopped for connection confirmation” for a failure. It is only waiting for a person to confirm, and it is not counted as a failure. It is not listed as a failure in “Action needed” either; it arrives as a review request (Connection Guide)
  • Thinking a run stops at its deadline. It does not. It is listed in “Action needed” and the owners are told, nothing more. To stop it, cancel that run
  • Thinking an owner can decide everything by approving in bulk from “Action needed”. Only review requests where you are named as a reviewer, or handed over to you as a delegate, can be decided. The other rows come back as “No permission”
  • Records and metrics — how the records work, and the thinking behind the figures
  • Review Guide — how to settle the review method
  • Execution Rules Guide — how much of an operation to allow
  • Design Guide — where to make the fixes when revisiting (goal, skills, drafts)
  • Settings reference — the periods for KPI and Insights, how many rows a Records (audit trail) CSV export holds, the deadline and stall hours, and how many rows “Action needed” lists