No freelancer or consultant enjoys it: filling in the timesheet on Friday afternoon. You crawl through your calendar, try to reconstruct what happened on Wednesday morning, and hope you forgot nothing. The result: undercharged time, inaccurate project margins, and procrastination until the end of the month.
Exact Online fully supports time tracking (per project, per employee, with cost rate and billability). But the screens are slow. With an AI assistant via MCP you log hours in natural language — sometimes faster than typing them into a spreadsheet.
Logging an hour in one sentence
The simplest workflow:
"Log 3 hours on project Renovation Bakery De Groot for today, activity Engineering, billable."
The AI calls the right MCP tool (items_time_create, combined with a project lookup), creates the entry and confirms. No menus, no clicking.
Bulk: a whole week in one prompt
For weekly entries:
"Log this week:
— Monday: 4h project Acme, 4h project Beta
— Tuesday: 8h project Beta
— Wednesday: 6h project Acme, 2h internal meeting
— Thursday: 8h project Gamma
— Friday: 4h project Gamma, 4h admin"
The AI parses your text, creates each line as a separate entry, verifies the projects exist and confirms the total. What used to take 20 minutes in Excel is done in one.
Recurring patterns
Work the same Monday structure every week? Save a pattern:
"Log my standard Monday structure for last Monday: 6h Acme, 2h meeting."
The AI remembers the structure within the conversation or via a reusable prompt in your prompt library. Minutes saved per week; hours over a year.
Forgot to log? Reconstruct
Friday afternoon, you forgot what you did on Wednesday. Ask the AI:
"Which projects were active in my calendar in week 28, and which hours have I already logged? Show me the difference."
The AI compares logged hours with — for example — your default working week or a connected calendar. Result: a tidy "missing hours" list instead of frantic searching.
Invoice billable hours directly
At month end:
"Per customer, give the billable hours from the past month with the calculated revenue."
Then:
"Create draft invoices for these hours per customer, ready for my review."
The AI chains projects_timecost_search and sales_invoices_create. See also month-end close with AI.
Validate against budget
Logging an hour on a project already 50 hours over budget? You want a signal for that:
"For which projects I logged hours on this week is the actual time more than 90% of budget?"
Immediate visibility on which project manager to call. Combine with our project margin guide.
Tips for effective AI time logging
- Use recognisable project names. "Project Acme" beats "project P2026-031" unless you know codes by heart.
- Make activity codes (Engineering, Meeting, Travel) explicit so the AI logs consistently.
- Log 5 minutes daily instead of 30 weekly. Faster, more accurate, no procrastination.
- For team managers: periodically ask "which employees did not log any hours on active projects this week?" — missing entries hurt margins.
Getting started
Time tracking is rarely anyone’s favourite, but essential for project margins and invoicing. By replacing "click in screens" with "type what you did", you dramatically lower the friction. Follow the MCP setup guide, check the prompt library, and pick the AI assistant that fits you.