My Agents

Your agents and their setups in one place. Run an agent, see its history, and choose when it runs and where its results go.

My agents lists the agents you built and the ones you got from the Store, each with its setups. Open a setup to run the agent on its Agent page, see what earlier runs gave you, and choose when it runs and where its results go, all without the canvas.

Your agents list

Open My agents under Automations in the app bar. The page shows how many agents and rooms you have.

Type in Search to find an agent or a room by name.
Show All agents, only Yours (the ones you built) or only those From the Store. From the Store shows no rooms.
Click Get agents to find more in the Agent Store.
When something needs you, a line above the cards says how many, for example "2 need you". Each item opens what it is about: a setup whose last run failed opens its History, memories waiting for your review open its Settings, and a room that waits for you (Your turn between rounds or A turn failed) opens the room. Click Show all to see every item, and Show less to fold them back.

Each agent has a card:

Its name, Yours or Store, and how many setups it has.
When it last ran, and All running fine or how many of its setups need you.
Its first setups, each with how it starts (its schedule, the app event that starts it, its seat in a room, or By hand) and when it last ran. Last run failed means its last run ended with an error; Not run yet means it hasn't run.
New setup makes another setup of the agent. See all opens the agent when it has more setups than the card shows.

Click the card or a setup to open it. Use Previous and Next under the cards when you have more agents than fit on one page.

Your rooms are listed under Rooms, and New room starts one.

The Agent page

The top of the page shows the agent's name and description. Click My Agents to go back to the list.

Built from the … workflow: shown to the agent's Builder. Opens the workflow on the canvas.
View listing: opens the agent's page in the Store.
Copy link or the share icons share the agent's Store page, when it is shared or public.
This setup is … in …: the room this setup has a seat in. Click it to open the room.

Below it, the SETUP bar shows which setup you are on. Pick another setup from it, or click New setup. Everything on the page is saved for that setup only; see Using an agent.

The page has these tabs: Run, Dashboards (only when the setup has dashboards), Settings and History.

When an agent has several actions, pick one under ACTION on the Run, Settings and History tabs. Each action has its own inputs, runs and settings.

Run an agent

On the Run tab, choose Conversation or Single run at the right of the tabs. Both run the same action with the same inputs; Using an agent explains the difference.

In Single run:

Under What you give it, fill in each input. Type text, drop a file or click Upload, click Record for audio, or pick one of the setup's tables, docs or dashboards. A hint under an input says what to put in.
Click Run (its Builder may have given the button another name).
What you get back shows the run as it goes, with Stop. Then it shows Done and the result, or This run failed and why. Click Try again to run it again. A run you stopped shows Stopped and Run again.

If the agent needs no inputs, just click Run. Until you run it, What you get back lists what the action gives you and where it sends its results.

A run carries on after you click Run: you can leave or close the page. Stop ends every run going in this setup, including another action's. Open the setup again while it runs and the page picks the run up. An action runs one run at a time: if it is already running, for example because a schedule started it, the page follows that run instead of starting another. If that run finished just as you clicked, the page says "The run that was going has just finished. Press Run again."

Where an input comes from

Each input shows where its value comes from. Click Change next to it to choose:

Ask each run: you fill it in when you run, and it is cleared after the run.
Set in advance: the value is saved with this setup and used on every run, including runs a schedule, a webhook or an app event starts.

An input set in advance can hold a queue of values. Click + Add to queue, or drop more files. Each run uses the next value, marked NEXT; the last one stays and is reused. An input marked Set by the Builder can't be changed.

Using under the inputs shows the setup's preferences. Click Change to edit them in Settings.

History

The History tab lists the action's runs in this setup, newest first. Each run shows what it gave back, what started it (By hand, Schedule, Your app for a webhook, App event or Room), when it finished and its credits. The line at the top gives all of the setup's runs and their credits. A run you stopped shows Stopped in grey, not as a failure.

Click a run to see, under What it gave you, what it was given and what it gave back. Click Show older runs to load more.

Settings

The Settings tab says "Changes save as you go." There is no Save button.

For this setup

Preferences: choices the Builder lets you make for this setup. Used every time the agent runs.
Memory: what this setup remembers. See Using an agent.
Apps and permissions: the apps the agent uses, each Connected or Not connected. The agent uses your own account for each app. Connect or Reconnect opens your App connections.
Secrets: values the agent's code needs, shown when it has any. Values are stored encrypted for this setup only, and nobody sees them.
- Needed: click Add value, type it and click Save. The agent can't run until you do.
- Provided by the Builder: works as it is. Click Use my own to use your own value, and Use the Builder's to go back.
- Set by the Builder: always the Builder's value; there is nothing to set.
- Your value: click Replace to change it, or Remove and confirm to take it off.

For each action

When it runs lists everything that starts the action. You can always run it by hand.

By hand: when you press Run.
On a schedule: pick the days (Every day, Weekdays, Mondays or Sundays) and a time, then switch it on. Switch it off to pause the schedule, or click Remove schedule to delete it. A one-time schedule shows as Once, with Remove schedule too.
From a webhook: lets your own systems start the action. The webhook panel under it has the keys, the address and which field of the call fills each input; see Webhook integration. Switch it off and calls are refused.
App events: each event that starts the action, such as a new email in your mail app, with its own switch. Click Add an app event to add one, and Close when you are done.

Runs that start on their own use the setup's saved inputs and preferences. A webhook call can also fill inputs from the fields it sends; inputs it doesn't fill keep their saved values.

Where results go lists the places the Builder lets you send results. Type an email address or a webhook address to turn that one on; clear it to turn it off. For Telegram, click Connect Telegram, press Start in the Neuron bot, then pick the chat; removing the chat turns it off. A place marked set by its Builder can't be changed.

If a run fails says what happens: either the Builder's error path runs, or nothing else runs and the error is logged in History.

Limits

File upload: up to 32 MB per file.
CSV upload for a table input: up to 25 MB.
Email recipients: up to 5 recipients.
Saved values per input: an input set in advance holds a limited number of values; adding more says how many it can hold.

Common problems

This agent can't run right now

The Run tab says why, for example because the agent was suspended or removed from the Store. See Agent Store.

Run is greyed out

The action is already running, or it can't run right now (the Run tab says why). In Conversation, Next and Run also wait for a required input. In Single run, a missing input is named when you click Run: "Add … first."

Run failed

This run failed shows what went wrong. Credits are used for any part of the agent that ran. Check the inputs and click Try again, or open the run in History.

A secret is Needed

Add a value for it under Secrets in Settings. The agent can't run without it.

An app is Not connected

Click Connect under Apps and permissions, connect the app, then come back. See App connections.