Can the AI change many records without asking?

No. Every change is a proposal: it is dry-run, shown to you, and applied only if you confirm within 30 minutes. Changing more than 5 rows needs bulk permission on the key, and more than 1,000 rows is refused.

What if the bulk change was a mistake?

Every applied change returns an undo, so the whole batch can be put back. Deleted records can be restored and a single record can be reverted to an earlier version.

Does it respect who can change what?

Yes. The AI acts as the signed-in person, so their permissions are the ceiling, and a key can be limited further to certain tables, rows or columns.

Bulk updates · Fieldproxy MCP

Change hundreds of records with one sentence

From Claude, ChatGPT or Copilot in Teams, on your live field service data, with a confirm step before anything changes.

Book a demo How it connects

Reply to Claude...

Fieldproxy

ClaudeChatGPTCopilot in Microsoft Teams

Illustration of Fieldproxy's MCP tools inside Claude, ChatGPT and Copilot. Names and data are examples.

Short answer

Bulk updates from ChatGPT, Claude or Copilot: how does it work?

Connected to Fieldproxy through MCP, Claude, ChatGPT or Copilot can make changes across many records at once: move a technician's jobs to the nearest qualified people and text the customers, raise every maintenance contract by a percentage, mark stale quotes lost, or push a region's jobs to another day. Each change is dry-run first and shown row by row, applied only when you confirm, and can be undone. More than 5 rows needs bulk permission and more than 1,000 is refused.

Last verified September 27, 2026

Real requests

What people ask, and the tools it calls

You ask	Fieldproxy tools	You get
“Mike is out sick: move his jobs to whoever is closest and qualified, and text the customers their new times”	technician_jobs, nearest_technicians, plan_job_distribution, apply_schedule, draft_sms, confirm_send	A re-plan plus customer texts, applied and sent on confirm
“Raise every active maintenance contract by 4% from January 1 and email each customer the new price”	find_contracts, propose_write, confirm_write, draft_email, confirm_send	A row-by-row dry run of every price, applied on confirm
“Mark every quote older than 60 days as lost, but keep the ones over $5,000 open”	quotes_to_follow_up, mark_quotes_lost, confirm_write	The stale quotes closed in one update, on confirm
“Push every Thursday job in the North region to Friday and keep the same technicians”	bulk_reschedule_jobs, confirm_write	A bulk reschedule, applied on confirm
“Set the priority to P1 on every open job at the three hospital sites”	find_jobs, propose_write, confirm_write	The matching jobs updated together
“That was wrong, undo it”	undo_write	The whole batch put back

The part nobody else does

Your AI can change the software, not just the data

Every other field service connector we checked (September 27, 2026) reads records and, at most, edits a few of them. Fieldproxy's MCP server can also build: new fields, screens, tables, automations and agents, made in a sandbox copy of your workspace and live only when you approve.

Add a field

"Add a warranty expiry date to every job." The column, the form field and the list column, in one ask.

Build a screen

"Build a subcontractor compliance screen that flags missing insurance." A new screen on your real data.

Wire an automation

"Email the service manager 30 days before a warranty runs out." Saved switched off until you turn it on.

Create an agent

Voice and text agents: validate, simulate against test conversations, then enable.

Add your own tools

Saved SQL tools such as "a customer's overdue invoices", which every AI connection in the workspace can then call by name.

Roll back anything

Every build is checked against your database before it saves, promoted from the sandbox when you approve, and can be rolled back.

Guardrails

Your AI proposes. A person confirms.

Built for operations with branches, teams and an IT review, not a single login.

Nothing changes on one call

Every write, send or action is a proposal the user confirms within 30 minutes. Writes can be undone. More than 5 rows needs bulk permission, and more than 1,000 is refused.

Keys scoped like API keys

Limit a key to certain tables, certain rows ("only the North region's jobs"), hidden columns, insert/update/delete, and an expiry from 1 hour to 90 days. Postgres enforces the row limits.

Builds save as drafts

Apps and automations are checked against the database before they save, new automations and agents are saved switched off, and public apps are never changed.

Sends are limited

Email goes only from the user's own mailbox, at most 20 sends an hour per key, inside the workspace's quiet hours. A sandbox never sends.

Everything is logged

Every call is written to an audit log with the tool, key, person and outcome. Administrators see every key and connected app, and revoking one applies on the next request.

FAQ

Questions

Can the AI change many records without asking?: No. Every change is a proposal: it is dry-run, shown to you, and applied only if you confirm within 30 minutes. Changing more than 5 rows needs bulk permission on the key, and more than 1,000 rows is refused.
What if the bulk change was a mistake?: Every applied change returns an undo, so the whole batch can be put back. Deleted records can be restored and a single record can be reverted to an earlier version.
Does it respect who can change what?: Yes. The AI acts as the signed-in person, so their permissions are the ceiling, and a key can be limited further to certain tables, rows or columns.

Bulk updates, from the chat

Book a demo