Jobber MCP now connects selected field-service records and workflows to ChatGPT and Claude, according to Jobber’s support documentation. The launch matters because it exposes useful business actions while deliberately withholding destructive and financially sensitive operations.
- Users can search requests, quotes, clients, properties, jobs and invoices.
- The connector can create or edit requests, clients and properties.
- It cannot delete records, alter schedules, or create jobs, quotes and invoices.
Jobber MCP turns business context into tools
Jobber is field-service management software used by home and commercial service businesses. Its new connector lets an authorized user ask an AI assistant for operational context, such as open requests or quotes awaiting a response, without copying records into a chat by hand.
The connector also supports narrowly scoped writes. Jobber says users can create or edit requests, clients and properties. Its examples include turning a work-order email into a request and producing a morning briefing from account data. The feature is available on Core, Connect, Grow and Plus plans, and the company says free ChatGPT and Claude accounts can connect.
The product is defined by what it cannot do
The most important design decision is the permission boundary. Jobber says the connector cannot create or edit jobs, quotes or invoices; it cannot change the schedule; and it cannot delete or archive any record. Those limits reduce the blast radius of a misunderstood instruction while still allowing common intake and lookup work.
This is distinct from giving an assistant browser access. Jobber explicitly says automated browser agents are blocked from signing in and directs customers to the supported connector. That separation makes authorization auditable and lets the application expose structured tools instead of relying on brittle screen automation.
Why vertical SaaS is becoming an agent surface
Field-service businesses store operational truth inside requests, properties, jobs and invoices. A general assistant becomes more useful when it can retrieve that context, but the connection also creates risk: the assistant should not silently alter money, commitments or worker schedules.
Jobber MCP offers a practical compromise. Read access covers the business objects needed for summaries and follow-up. Write access stops at intake-style records, where a human can continue the workflow. The product’s value will depend less on how many tools it exposes than on whether permissions, confirmations and activity records remain clear.
| Capability | Status |
|---|---|
| Search operational records | Supported |
| Create requests and clients | Supported |
| Change schedules or invoices | Not supported |
| Delete records | Not supported |
The Lapaas view
Everyone else is reporting another MCP launch; we are explaining the permission model that makes it deployable. The competitive advantage is not merely connecting a chatbot. It is translating a vertical application’s roles and objects into tools that remain useful without becoming dangerously broad.
That pattern also appears in our coverage of Google Home’s MCP control plane and Amazon’s Claude seller plugin. Each product is ultimately judged by which actions it permits, which it blocks and how clearly the user can see the difference.
Jobber MCP shows that useful business agents do not need unrestricted access: narrow read and write tools can automate intake while leaving money, scheduling and deletion behind explicit product boundaries.
Frequently asked questions
What does Jobber MCP do?
It gives authorized ChatGPT and Claude connectors structured access to selected Jobber records and a limited set of create or edit actions.
Can an AI agent delete Jobber records?
No. Jobber says the connector cannot delete or archive records.
Can it change schedules or invoices?
No. The current connector cannot edit schedules, jobs, quotes or invoices.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
