Start from a real task event
Call enrichment when a known record is created, opened, assigned or needed for a specific decision.
Controlled MCP enrichment workflows
Add waterfall contact enrichment to AI agent workflows.
Let an agent request a work email or business phone number for the known professional in its task. TargetWise routes each field through eligible specialist sources and returns one result while the host controls tool access, call budgets and downstream action.
Built for paginated search and selective enrichment, with a clear unavailable-data branch.
Illustrative agent workflow branch
IllustrativeRequestFind a work email for this known professionalOne profile supplied · contact tool selected
TargetWise routeOne entity per lookup
IdentityPerson and company resolved
Work emailReturned
—Automatic outreachApproval required
15+ sourcesavailable for contact waterfall routing
11 toolssearch · enrich · field lookup
25maximum candidates per search page
Explicitmatched, partial or not found
Agent versus workflow
Reasoning can be flexible. Data operations should be predictable.
An agent can decide that more context would help; the workflow should decide whether the call is allowed and how each outcome is handled.
This division makes the system easier to test and operate. The model identifies the need for context or synthesizes a returned result. Deterministic logic validates identifiers, selects the operation and allowlisted contact fields, filters company-profile fields, limits calls and decides whether a result can be merged or requires review.
TargetWise remains a focused retrieval step. It can search a page of up to 25 candidates, but it does not scan in the background, auto-enrich every result, write to a CRM or send a message.
Give each task a controlled enrichment path.Set call limits, require new identity evidence before rerunning unresolved requests, and keep any write or communication behind an explicit decision boundary.
Workflow controls
Design the result handling before the first call runs.
A useful enrichment workflow explains what happens on success, partial data, no match and operational failure.
Validate the available identifiers
Check that the workflow has enough context to distinguish the intended person or company.
Use a data allowlist
Select only needed contact fields and filter which returned company-profile fields this branch may consume.
Handle returned and unavailable data
Use the documented response rather than interpreting an empty value or inventing a field.
Keep retries controlled
Require new identity evidence before rerunning an unresolved request and follow documented error handling.
Separate retrieval from writes
Apply merge rules and approval before updating a system or using returned contact data downstream.
Reference flow
One waterfall request, four controlled stages.
The same pattern works whether REST or MCP delivers the enrichment step.
Trigger and identify
Start on one task, inspect existing fields and choose the intended entity.
Select and call
Allowlist the operation and data scope, then make one focused TargetWise request.
Evaluate the result
Apply clear rules to returned and unavailable data.
Review and record
Approve consequential actions and log what the workflow used and decided.
Operational discipline
Prevent loops, silent overwrites and hidden actions.
Agent systems become expensive and hard to trust when retries and tool calls are open-ended. Give each enrichment branch clear call limits and an unavailable-data path.
Preserve customer-provided or recently confirmed values unless an explicit merge rule allows a change. Keep CRM writes, exports and communications visible as separate steps that can require approval.
- Call limits for each workflow
- New identity evidence before rerunning unresolved requests
- Documented recovery for operational errors
- Field-level merge rules before overwriting
- Approval before writes, exports or communications
Branch logic
A predictable path for every outcome
1Data returned — apply acceptance rules
2Some data returned — use only required fields
3Unavailable — stop or request stronger identity
4Operational error — follow the recovery policy
Workflow patterns
Enrich at the moment a real decision needs context.
Avoid scheduled whole-database refreshes and open-ended agent loops. Trigger enrichment from a meaningful known-record event.
Find a work email before routing a known lead.
Request the missing field for the submitted professional, then return the result to the customer’s assignment logic.
Request a business phone during record review.
Call the contact tool for the record currently open and keep any write behind the CRM merge rule.
Discover or enrich the account deliberately.
Use filtered company search when candidates are needed, then enrich only the account selected for the brief.
Expose only the required enrichment tool.
Let the host allowlist one TargetWise MCP action and cap repeated calls for the task.
Frequently asked questions
What to know before you build.
The most reliable workflow is explicit about identity, field scope, terminal outcomes, retries, merge rules and human approval.
Build one workflow first
Use the Free plan to test the full branch—not only successful matches.
Include known matches, difficult records and expected no-match cases so you can verify the controls that matter before production use.