Research
Hand off a question and let it dig.
Point an agent at last year's numbers and ask which Black Friday promo to run again. It comes back with the comparison and the reasoning behind the pick.
Agents
Give an agent a goal. It works until the job is done.
It plans the steps itself and runs them on data Switchboard's already joined across your tools. Or set a rule instead.
Marketing Budget Recommendation
GoalHit target ROAS across channels
ScheduleEvery Monday · 6:00 AM
Run · Mon 6:00 AM Complete
- 6:00Pulled spend and performance
- 6:00Read seasonality context
- 6:01Compared channel efficiency
- 6:02Drafted budget shifts
- 6:02Posted to #marketing
Runs on data from every tool.
Agents start from the data Switchboard has already joined: orders, stock, ad spend, support. There's nothing to wire together first.
Writes the result back to your tools.
An agent can place an order in your inventory system, update fields in HubSpot, or send a summary to Slack or email.
Runs on a schedule or on demand.
Run one every morning, every Monday, or once for a single job. Each run is logged, and you can pause it any time.
Give it a goal, or set it a rule.
Recommend
Run the weekly analysis and recommend the move.
Every Monday, review ad performance and recommend how to shift next week's budget.
Sync
Keep the tools your team lives in current.
Push each account's health score and lifetime value into HubSpot and Klaviyo, recalculated nightly. The numbers your team sees in the CRM match the ones in Switchboard.
Act
Set a rule and let it run, the same way every time.
Fill in a new lead's company details from your ad and form data the moment it lands. No AI making the judgment call.
FAQ
What can an agent do on its own?
Only the job you gave it. An agent works toward its goal with the data you pointed it at and the destinations you set. Rule-based jobs are narrower still: the rule is the whole job.
How do I know what a run did?
Every run keeps a log: what it read, what it decided, and what it sent where. You can check any run after the fact, and pause the agent any time.
What's the difference between an agent and an automation?
An automation is a fixed rule: when this happens, do that, exactly as written. An agent takes a goal and figures out the steps itself. Both run on the same joined data; you choose how much latitude the job deserves.
Can it act in a tool Switchboard isn't connected to?
Agents act through your connected tools. If your workflow needs to touch one we don't cover yet, bring it to us.
Bring us the workflow.
Describe the Monday process that eats your morning. Our team maps it against your data and builds the agent with you: the goal, what it reads, which tools it touches. You review what it produces.
Book a demoGet started
The work you're tired of doing by hand.
Book a demo. We'll show you how Switchboard would run it.