Native integration
Published only after a supported, maintained connector is verified.
Claire’s API Call, variables, branching and workflow actions support configurable connections. A system name never means a native integration unless Claire has verified and published that status.
Claire can be configured to work around authorized systems and APIs. Integration status is validated during implementation.
Published only after a supported, maintained connector is verified.
A supported API connection built for an agreed workflow.
Connection feasibility depends on API access, credentials and implementation scope.
Named plainly when a requested connection is not currently supported.
Secrets belong in controlled credential storage, not page content or prompts.
Unavailable actions route context to a person rather than pretending completion.
Decide where Claire executes, where a person approves, and which exceptions require takeover.
Configured retrieval, preparation, communication and actions.
Judgment, approval, relationships and consequential decisions.
Before Claire reads or writes anything, implementation must establish what the target system permits, which records matter, how identity is matched, what happens when the system is unavailable and which action requires a person.
Name the exact read or write: retrieve a customer, check a schedule, create a work order, update a load, attach a document or record an outcome.
Confirm the customer’s edition, API availability, authentication method, scopes, rate limits and any vendor approval process. A public API does not guarantee customer access.
Decide how callers, contacts, accounts, properties, loads, claims and assets are matched—and which fields are required before an action is valid.
Limit each connection to the records and operations the workflow needs. Consequential or ambiguous actions can route to a human with context instead.
Handle timeouts, missing records, duplicate matches, validation errors and unavailable systems explicitly. Never tell the customer the work is complete when the write failed.
Test the conversation, data mapping, system result, outbound confirmation and exception path together using approved non-sensitive test data.
The current named-system pages describe possible configurable API workflows. They do not assert a native connector, marketplace listing, vendor partnership, certification or existing customer deployment.
Bring the request, systems, policies and handoff points.
The most commercially useful integration is not always the best-known system. It is the connection that unlocks several high-value workflows for the customers Claire is positioned to serve. Prioritization should consider customer penetration, available access, implementation feasibility, workflows unlocked, exception behavior, and revenue leverage.
Which identity, customer, job, appointment, load, unit, policy, claim, inventory, or ticket must Claire retrieve—and how is the correct record matched?
Must Claire create, update, attach, schedule, cancel, dispatch, or record an outcome? Which writes require validation or approval?
Does the workflow wait for a webhook, human decision, vendor acknowledgement, carrier event, document, or scheduled time before resuming?
What happens on timeout, rate limit, duplicate match, validation error, partial completion, unavailable system, or revoked credential?
Every published integration page states the connection status precisely: native, API-capable, configurable, or not currently available. A vendor name alone is not evidence that Claire can access a customer’s edition or perform a particular write.
The connected-system demonstration should show more than a logo: the object retrieved, proposed or completed action, permission boundary, response, error path, and system of record. Where live access is unavailable, label the screen as a workflow example and explain that customer-specific integration validation is required.