Build custom tools for bots.
Test once on a real lead. Reuse in branches, stop rules, AI steps, and AI SDR flows.
Save reusable API tools once, test them on a real lead, then reuse them anywhere the workflow needs proof. Campaigns, AI replies, and AI SDRs can all call the same custom tool instead of rebuilding it every time.
Last updated:
Build the check.
Define the method, URL, headers, body, and response mapping one time for the whole workspace.
Try it on a lead.
Use a real lead when you test so the API call reflects actual record data instead of mock placeholders.
Call the same tool everywhere.
The same tool can show up inside AI steps, stop rules, branch decisions, and AI SDR logic without rework.
Keep a default path.
If the API call fails during a branch, the lead can still fall through the default path instead of breaking the campaign.
API check
Test once. Reuse everywhere.
Tool test running for lead: Sarah Chen · Endpoint: /credit-status
Mapped response: data.status = approved
Branch outcome: approved leads continue to the AI offer path.
Branch logic
Fallback path remains attached if the tool errors or returns nothing.
Default path
The same API tool can support campaigns, stop logic, and AI-driven branches.
Run the tool against actual lead data before you trust it in production.
Turn a raw API payload into a field the campaign can actually use.
If the tool cannot decide, the branch can still continue on a defined fallback path.
External logic. Internalized.
Save the API logic once at the workspace level, then reuse it wherever campaigns, replies, and AI bots need more context.
Pick a lead before you run the tool so placeholders, headers, and mappings reflect the data your workflow will actually see.
Use API outcomes to choose the next path instead of hoping the CRM already stores every field you need.
Ask one more system for approval before the message goes out and stop the flow if reality changed.
AI-driven steps can use the returned data so copy, offers, and next moves reflect the external truth instead of guessing.
When the tool errors, your workflow can still fall through a default branch instead of dying mid-stream.
Workflow
How teams use it.
Save the tool. Test it. Reuse it wherever a branch needs one more fact.
Define the endpoint.
Set the method, URL, headers, request body, and the response field you care about.
Run a live test.
Choose a real lead and confirm the tool returns the value your workflow expects.
Attach it to logic.
Drop the tool into a stop condition, a branch, or an AI step that needs the answer.
Keep the fallback.
If the call fails, the default path still protects the campaign from breaking.
Related
Works with.
Ask the stack first.
When a message depends on inventory, credit, lifecycle, or any other outside system, your bot should be able to call the tool before it sends.