Summary
The "Agents" Module is where you see what used to be called your Automations and your Integrations (which were in the App Center in the previous version of Planhat)
It shows you how each Agent has been performing, and surfaces anything waiting on you
For each Agent, you can see who created it, how "healthy" it's been over the last 7 days, when it last ran, how many times it has run, and whether it's switched on
Two elements sit immediately above the Agents list: "Needs attention" for approvals waiting on you, and runs that failed; and "In progress" for runs happening right now
From the Agents Module you can create a new Agent, switch it on or off, open it, copy it, view its logs, delete it, and approve or reject steps
Who is this article for?
Anyone using Planhat OS who wants to see what's running in their tenant
It's particularly relevant for Planhat Users who own Agents, and for anyone who gets Agent run approval requests
It's also relevant to Planhat Admins (e.g. Ops) responsible for managing Agents overall
Article contents
What is the Agents Module?
The Agents Module gives you an overview into automated processes you own or have been set up by your colleagues.
Here you can view how your Agents have been performing, and see if there is anything that needs your attention, allowing you to triage and resolve issues quickly.
The Agents Module contains what used to be called your Automations, and also your Integrations (e.g. for Salesforce, Intercom and Zendesk etc.), which were in the App Center in the previous version of Planhat.
Everyone can view the Agents Module. What each User can do within it depends on their Role permissions.
📌 Important to note
The old "Automations" have been renamed to "Agents" in Planhat OS
OAuth Clients, Private Apps (Service Accounts) and Connections, which were all previously in the App Center, are now in the main "Settings" System Admin Module, rather than being in the Agents Module
Similarly, Project Templates and Sequence Templates (both Workflow Templates), other ways to automate processes, are now in their own "Tools", rather than being in the Agents Module
The Agents Module also does not include the main, in-built, general purpose Planhat AI Agent that you chat to in Sessions
Click the image to view it enlarged
The Agents Module has five elements, as shown in the screenshot above. Top to bottom, they are:
"+ New agent" - the button in the top right for creating an Agent
"Analytics" - how much your Agents are running (number of runs), how often they succeed (success rate), and what they're costing in credits
"Needs Attention" - approvals waiting on you, and runs that failed
"In Progress" - runs happening currently
"Agents" - the full list
Next in this article, we will take a more detailed look at each of these in turn.
How to create a new Agent
Let's go through how you can create a new Agent (formerly called an Automation) via the Agents Module.
Click "+ New agent" in the top right. You can choose between two options:
Option | Use this option when ... |
"Create with AI" - describe it in chat | You know what you want the Agent to do, but not how to build it |
"Create in Canvas" - start from scratch or a template, via the Agents Library (formerly called the Apps Library) | You want to build the Agent yourself, or a template already exists for the use case |
Create with AI
Click "+ New agent", then "Create with AI"
Describe what you want the Agent to do, in your own words: what should trigger it, and what should happen
Planhat builds the Agent from your description. Review what's been produced before you switch it on
🚀 Tip
Be specific about the trigger. For example, instead of saying "When a Deal closes", say "When a Deal Stage is updated to Closed Won".
Examples of prompts
Create a weekly email prep agent for me. Manual trigger. When triggered, you should scan through conversations and tasks from the past week. Distill key info to share with the customer. Draft a weekly summary email I can share with the customer. Enable HITL for email step.
Click the image to view it enlarged
Build a renewal agent for me, remind customer 90 days before renewal, send no more than 5 emails in total. You should use AI to summarize key values achieved during their subscription, and reference evidence such as metrics, etc. You should also link the renewal with their company status (using external research). Exit the flow as soon as the customer replies.
Click the image to view it enlarged
Build me a risk signal detection agent that detects the following types of risk from the conversation. Trigger this agent automatically whenever a conversation is created. Store the risk category, risk analysis, risk level in conversation data model custom fields. Risk types: Churn risk, Competitor mention, Escalation Expansion, Renewal hesitation, Product issue, Onboarding struggle, Champion change, Budget concern.
Click the image to view it enlarged
Create in Canvas
Click "+ New agent", then "Create in Canvas", to open up the Agents Library (what was called the Apps Library in the previous version of Planhat)
Click the image to view it enlarged
Either select a Templated Agent (Templated Automation) ...
Note that these are still customizable, despite being templates
... or select "Custom agent" to build from scratch
Build and activate the Agent (Automation) as normal - see our existing Automations articles for further details
📌 Important to note
Some of the access here is controlled by Role permissions. If you don't have the ability to create a new Agent and you think you should, speak with your organization's Planhat admin, or your Planhat contacts.
🚀 Tip
The Agents Module is also where you will find native Integrations (e.g. for Salesforce, HubSpot, Zendesk, Jira, Snowflake and so on), at the time of writing. To add a new Integration, in the Agents Module, click "+ New agent" and then "Create in Canvas" to open the Agents Library. Then, use the search box in the top right to quickly find the Integration you're looking for; or, you can browse to it by using the category tabs in the bottom left (under "Integrations").
Analytics
The Analytics part of the Agents Module shows how your Agents are performing across the whole tenant, rather than one at a time.
Click the image to view it enlarged
There are three built-in metrics/KPIs at the top, and for each of these, if you click on it, it will show that data in the graph below. The graphs show you the trend, so you can see whether things are getting better or worse rather than only where they stand today.
"Runs" - how many times your Agents ran over the last 7 days
"Success rate" - the share of those runs that completed without failing
"AI credits" - credit consumption, which you can break down by "AI", "Automation" or "Email"
🚀 Tip
Use Analytics to spot a pattern, and use the Agents list to find the cause.
A drop in success rate without a matching drop in runs usually means one Agent started failing, rather than everything degrading.
You can look at the "7D health" column in the Agents list to see which Agent is responsible.
Needs Attention
Approvals and failures are gathered in one place and can be actioned inline - you don't need to open the Agent to approve a step.
Two main types of item appear in the "Needs Attention" part of the Agents Module:
Approvals - waiting for a decision
This is related to "Human-in-the-Loop (HITL)" in Agents (Automations), which we have a separate article on
You can approve or reject the paused run, directly from the row - see the "x" and checkmark (tick) in the bottom right of the screenshot below
This shows your own notifications - what you personally were notified about, not everything outstanding across the tenant (your organization's Planhat environment)
Failures - runs that errored
Open the run to see what happened, or mark it as read to clear it
Click the image to view it enlarged
Up to 10 items show here, and you can expand to show the rest.
When there's nothing outstanding, it reads "Nothing needs attention".
In Progress
"In Progress" displays runs that are currently running or paused, each showing how long it's been going, updating live. You can click a run to open up its details on the right-hand side (example shown in the screenshot below).
A run that's been going far longer than usual is the clearest sign something has stalled - for example, an Agent (Automation) that has been waiting for a long time for a Human-in-the-Loop (HITL) approval.
Click the image to view it enlarged
Agents list
Here you can see all of the Agents (Automations/Integrations - not the main Planhat AI Agent you chat to, or Project/Sequence Workflows) in your tenant.
There are two tabs: "Mine" for the Agents that you have set up, and "All" for everything in your tenant (so including Agents set up by your colleagues).
Each row shows:
Column | What it tells you |
Name | The Agent's name and icon |
Created by | The User who created it |
7D health | The share of runs in the last 7 days that succeeded |
Last run | How long ago it last ran |
Runs | Total times it has been triggered |
Enabled | Whether it's switched on or off. Click the toggle to enable/disable |
The list shows the 10 most recently created Agents. To see more, put your mouse to the bottom of the list, and click the "More" that appears.
You can also use the search box to find a specific Agent:
🚀 Tip
7-day health is blank when an Agent hasn't run in the last 7 days.
When health is shown: green means every run succeeded, amber means some failed, red means all of them failed.
Health over the last 7 days means that any broken Agent is visible in the list, rather than only in a notification. Therefore, you notice failures, without having to actively look for them.
Agent actions
When you mouse over an Agent in the list, it will show an ellipsis symbol (3 dots). Click on this to open up a menu with the following options:
Action | What it does |
Copy link | Copies a direct link to the Agent |
View logs | Opens the logs, already filtered to this Agent |
Copy | Makes a duplicate that you can modify |
Delete | Removes the Agent permanently, after confirmation |
You can also switch an Agent on or off via the "Enabled" toggle, without opening the ellipsis menu.
📌 Important to note
Deleting an Agent cannot be undone. If you just want to stop it running, switch the "Enabled" toggle off instead.
Opening an Agent to view details
You can click an (Automation-type) Agent's row to open up further details on the right-hand side.
Click the image to view it enlarged
You'll see:
"Total Runs", "7D [7-day] Health", and "AI credits - 30d [used in the last 30 days]"
"Overview"
Who created the Agent, when, who last updated it, and when it last ran
"Canvas"
Here you can see the Agent (Automation) steps in a flow chart
Click the pencil icon to be able to edit the steps
"Monitor"
Enable/disable "failed" and "disabled" notifications
View the daily failure threshold
"Approvals"
Related to Human-in-the-Loop (HITL)
Here you'll see anything pending, with options to approve or reject
"Recent Runs"
The Agent's run history
"Related Sessions"
The Sessions connected to this Agent
You can see any chats where the Agent was originated or where it was discussed
🚀 Tip
"Related Sessions" is the quickest way to view the reasoning behind an Agent - particularly one built via "Create with AI", where the original description is the closest thing to a specification.
Frequently asked questions (FAQs)
Where did the App Center go?
The App Center has been deprecated
What was called Automations are now Agents, and they can be found in the Agents Module
Native Integrations are also located in the Agents Module
OAuth Clients, Private Apps (Service Accounts) and Connections have moved to "Settings" within the "System Admin" menu
Why can't I see a failure my colleague was notified about?
"Needs Attention" is built from your own notifications (Agent actions where you personally are notified); these notifications are not shared with your whole tenant/organization
Why can't I create an Agent?
You may be lacking Automation permissions
What's the difference between the "Create with AI" and "Create in Canvas"?
"Describe it" versus "build it"
Both produce the same kind of Agent (previously called an Automation)
Can I edit an Agent that was created with AI?
Yes, you can edit an Agent after it's created by AI