AI and assistants
Revenue agents and automations
The five named agents that work your pipeline in the background, and the workspace automations you define yourself.
Updated August 2026
Where the assistant waits to be asked, agents run on their own. They watch your data, notice things, and put work in front of you. They do not act unilaterally on anything that leaves the account.
The five agents
| Agent | What it watches |
|---|---|
| Acquisition | Runs your saved lead searches, scouts new listings, and queues buyer-matched opportunities for review. |
| Deals | Keeps active deals moving. Drafts follow-ups and renewal outreach. |
| Due Diligence | Tracks deals toward closing, flags diligence gaps and stalls, and advances stages when criteria are clearly met. |
| Financing | Watches the capital side of deals in progress. |
| Lease Admin | Watches lease dates, options, and reconciliation obligations after a deal is done. |
How work reaches you
Agents produce action cards: a proposed piece of work with the reasoning and the underlying data attached. You approve, edit, or dismiss. Nothing that touches a client, whether an email, a message, or a calendar invite, executes without that approval.
Low-risk internal actions are the exception. Creating a task or logging an interaction can run automatically, and every one is recorded in an audit trail you can review.
Running an agent on demand
Agents run on a schedule, but you can trigger one manually when you want a fresh pass, before a pipeline review for instance. The activity feed shows what each agent did and when.
Workspace automations
Agents are ours; automations are yours. In the workspace builder you describe a recurring instruction in plain English and it becomes a reviewable recipe with a trigger, a condition, and a set of actions.
Recipes are created paused. You read what it will do and turn it on deliberately. Supported triggers today cover new contacts, new buyer call notes, deals with no activity for a set period, and a daily scheduled check.
The same boundary applies: task creation and interaction logging can run automatically; email drafts and calendar actions stay approval-bound.
Turning it down
If the agents are producing more than you want to review, you can disable individual agents or narrow which notification categories reach you. An agent you have turned off stops running rather than running quietly.
Usage
An agent step that needs the assistant to think, such as drafting a follow-up, counts as an AI message. Deterministic steps do not.