What are Benchling Agents?
Agents let you automate multi-step workflows in Benchling. This article walks through the full lifecycle: turning the capability on, setting up an agent, using and sharing it, how it differs from regular Benchling chat, and how credit limits apply.
Note: While in Preview, Agents do not consume credits.
Agents vs Chat
Agents and Chat are both part of Benchling AI, but they're built for different things:
Chat is the default Benchling agent: get AI-powered assistance by asking questions or having it take actions on your behalf within the conversation.
Agents are set up once and can be reused, shared, and run repeatedly on a trigger, with their own defined scope and their own identity independent of who's currently interacting with them.
Get started with example Agents
- To help you get started, example Agents can be found at the top of the Agent configuration page.
- Click one of the examples to pre-populate the agent configuration, make any adjustments as needed.
- Click Save.
Agent types
When you create an agent, you'll choose between two types:
Agent that runs with your permissions
The agent can only see and act on what you, the creator, has access to.
Good fit for individual workflows you want to automate for yourself.
Can be shared with others in a read-only capacity.
Agent that runs with its own permissions
The agent can be added to a team or project to access data - this must be approved by admins before agents gain any data access.
Useful for:
shared, team-level automations that need to keep running even if the person who set it up changes roles or leaves.
setting up an automation that has a limited set of permissions
Can be shared with others in read, write and admin capacities.
Setting up an Agent
Agents can be configured via Chat or manually via a form.
Configure an agent via Chat
- Start a new chat
- Describe the agent you want setup
- Confirm the proposed plan
- Optionally, review your new agent or share it with collaborators
Configure an agent manually
Go to AI Chat Settings > Agents (a modal will open)
Click Add
Specify Name, Description, Instructions.
Think of Instructions as a normal chat prompt. It will be sent to the agent when triggered.
Configure the Trigger type (time based or event based).
For event-based triggers, we currently support a set of events for Study, Entry, Run, Analysis, File, Workflow Task, Custom Entity and DNA Sequence and RNA Sequence.
Note: Depending on the data type selected, the supported events will populate below (eg for Entry, an agent can only be triggered on entry creation - we are working on building out support for more events. If there are particular events you’d like to see, please let us know!)
For time-based triggers, days of the week and time of day can be specified. Along with how often you want this agent to run.
Note: With this configuration, the Agent would run every Monday morning at 8am PDT.
Choose the agent's permissions
Select Uses mine if your agent should run with your permissions. (Tip: good for personal agents you don’t intend to share or have others use)
Select Has its own if your agent should run with its own permissions, separate from any individual user. (Tip: this is a good choice for shared agents that other users need to be able to interact with.)
Choose the Agent mode
In Default, the agent can research and draft, but will ask for approval before editing Benchling objects.
In Full actions, the agent will be able to edit or create Benchling objects and share its outputs with teammates.
Optionally, in Full actions mode a destination folder can be selected to limit where the agent can edit or create objects. Before making edits outside of this folder, the agent will prompt for your approval.
Add collaborators
If other users should be able to view and interact with your agents, add them as collaborators with the relevant policy
Click Save. The Agent will now fire automatically based on the trigger, using the same permissions as the user who configured it.
Requesting team membership for your agent
Agents with their own permissions have no data access by default, so they require an extra step for requesting data access.
It is recommended to add your agent to a team to manage its data access.
To do this, add a team from the Grant data access page. The membership will show as pending until an admin approves it.
Note: once an agent has access to specific data, any user added as a collaborator will gain access to that data (regardless of whether they have access on their own)
Managing an agent
Once saved, an agent can be paused (to stop it temporarily, with the option to resume later) or archived (if you no longer need it).
Active and Paused agents can be viewed in the Agent listing view. In the image below, the first agent is paused and the second is active.
Using and interacting with your agent
Once set up, an agent runs automatically based on its trigger. You don't have to kick it off yourself, unless you want to chat directly with it.
Agent chats show up in your chat history alongside your regular chats. A robot icon distinguishes an agent chat from one you started with Chat
Assuming you have sufficient permissions, you can continue any agent-triggered chat with follow-up messages, just like a regular chat.
To get notified when an agent runs, enable agent notification emails in your settings.
Event-based vs. schedule-based triggers
| Event based | Time based | |
| Fires when | A specific event happens (e.g., study phase changes, an entry is created) | A fixed cadence (e.g., daily, weekly, every Monday at 9am) |
| Best for | Reactive workflows - respond the moment something changes | Proactive/routine workflows - recurring checks, summaries, reports |
| Example | "When a study phase changes from design to execution, summarize it and flag anomalies" | "Every Monday morning, pull all approved entries, review their contents against our company entry guidelines skill, and flag any that don’t follow those standards." |
Tenant Admin controls for Agents
Any user can create a personal agent, which runs with their own permissions. Tenant admins control who can create agents that run with dedicated permissions.
This is governed by a toggle in the AI Settings of the Tenant Admin Console.
Admins can enable this capability tenant-wide, or scope it to specific users.
Once enabled, eligible users will see the option to create an agent that uses its own permission
Agent Sharing
Agents can be shared with other users so your entire team can benefit from them.
Users must have the Admin policy on an agent to share it (by default only the creator and tenant admins have this policy)
Shared agents appear in the collaborator's Shared with me tab in AI Chat Settings. The agent's runs will also appear automatically in the Recent chats of all collaborators.
Agent collaborators can be updated at any time from the agent's settings.
Available access policies:
View agent - Can view agent configuration and runs
Interact with agent - View agent + start new chats and send followups
Edit agent - Interact with agent + edit agent configuration, pause and archive the agent
Manage agent - Edit agent + manage collaborators
Usage tracking and limits
While in Preview, Agents will not consume credits, but their usage can be seen in the AI usage chart by checking “include usage that doesn’t deduct from credit limits”. Agent usage counts toward the credits of the account it runs as. This means:
Agent running with your permissions: adds to your credit usage.
Agent running with its own permissions: adds to the agent's credit usage.
Agents is a separate product type in the AI usage chart
Agents running with their own permissions can be searched for individually in the "User or Agent" field. This lets admins view how many credits a specific agent has consumed.
Agents running with the creator's permissions are not searchable here, since they run using the creator's identity, not a separate one.
Users are searchable in the "User or Agent" field.
Agents running with the creator's permissions contribute to the creator's credit usage.
Filter down to a user to see how much of a their credit usage comes from Agents, Chat, or other sources.
Usage limits
Regardless of credit consumption, automated agent runs are also subject to a rate limit. An automated run is one triggered by a time-based or event-based trigger (as opposed to a direct chat).
Across the whole tenant, Benchling AI supports up to 20 automated agent runs per rolling 10-minute window. Runs beyond that threshold are skipped, and affected users are notified by email and in-product.
Note: this limit doesn't apply to direct chats with an agent.