Short answer
Yes, when the software lets it and puts a person in the loop. Connected to Fieldproxy through MCP, ChatGPT, Claude or Copilot can change hundreds of jobs, quotes, contracts or customers from one request, and import up to 1,000 spreadsheet rows at a time. Every change is dry-run first and shown row by row with before and after values, applied only when you confirm, and can be undone as one batch.
What to ask
- "Mike is out sick. Move his jobs to whoever is closest and qualified, and text the customers their new times."
- "Raise every active maintenance contract by 4% from January 1 and email each customer the new price."
- "Mark every quote older than 60 days as lost, but keep the ones over $5,000 open."
- "Import this spreadsheet of 400 new sites."
- "That was wrong, undo it."
How it works
The assistant prepares the change with propose_write, import_records, bulk_reschedule_jobs or mark_quotes_lost. Each one is a dry run: you see how many rows it touches, sample rows and the before and after values. Nothing is saved until confirm_write, and several statements can be applied together, all or nothing.
Changing more than 5 rows needs bulk permission on the key, and more than 1,000 rows is refused. Every applied change returns an undo, deleted records can be restored, and one record can be reverted to an earlier version.
Why this is harder than reading data
Many AI connectors for business software mostly read records. A bulk change is where a wrong guess costs real money, so it needs a dry run, a person's yes, permissions that follow the signed-in user, and a way back. Those are the parts to check before you let any AI change your data.
Fieldproxy facts on this page come from the Fieldproxy MCP reference. Competitor details are from each vendor's own documentation as of September 27, 2026; see the comparison table and sources.