These are the prompts from our second MCP webinar, where we took a fully planned job and turned it into real work for real people, then threw the usual curveballs at it: a client delay, unexpected leave, and a new job landing.
🕰️ Est. Time: 10 minutes
🎯 Goal: Use five prompts to assign your team, build a realistic schedule and keep it on track when plans change
🔑 Requirements: Streamtime connected to your AI assistant through the MCP, a job planned by role with dated items, and the permissions listed under each prompt
Prefer to watch? The full webinar replay is here.
Each prompt below has a timestamp, so you can skip straight to it.
✨ Still haven't connected to the MCP?: you can find the steps to do so here.
Begin by going through the prompts from part one, where we built the job these prompts use.
Before you start
It works with your permissions. The MCP can only do what you can already do in Streamtime.
Swap anything in [square brackets] for your own details: job names, people, dates.
Give everyone a role in Streamtime. Most of these prompts match people to work by role.
Keep working hours up to date. Availability is worked out from each person's hours and what they're already booked on, so the answers are only as good as that data.
Workflow Prompts
Prompt one: swap roles for people
⏱️ Timestamp: 3:32
Replace every role on the job with an actual team member. Your assistant checks capacity across each item's dates before picking someone, splits heavy loads by type of work so each person owns related tasks, and makes sure nobody joins late in the job with no context. It shows you its proposed assignments and reasoning first, and only makes changes once you've approved them.
Prompt:
On the [job name] job, every task is currently planned by role. Swap each role for real team members. For each role, check who has capacity across the task dates, and pick the best fit. Only use actual team members, not placeholder or freelancer users.
If one person would end up carrying a heavy load across the job, split the work between two people. Split it by type of work, so each person owns related tasks, rather than dividing individual tasks down the middle. Nobody should join late in the job with no context. If someone takes on later-phase work, give them earlier tasks that feed into it, so they've been part of the job from the start.
Show me the proposed assignments and your reasoning first, then make the changes once I've approved them.
🔧 Set up:
Plan the job by role, with dates on every item. Capacity is checked across each item's dates, so undated items can't be matched properly.
Make placeholder and freelancer users easy to spot, for example with clear names, so your assistant can leave them out.
Keep the approval step in the prompt. It's your chance to check the reasoning before anything changes.
🔑 Permissions:
Edit Jobs: to assign team members to the job's items.
View Team To Do: so your assistant can see what everyone's already booked on when it checks capacity.
Prompt two: build the schedule
⏱️ Timestamp: 9:15
With people assigned, your assistant creates To Dos for each person across their item dates, working around what they're already booked on. A daily cap keeps it realistic: nobody wants eight hours on one project, and it leaves room for everything else in their week. You get a week-by-week summary at the end, with anyone over capacity flagged.
Prompt:
Now schedule the [job name] job. Create to-dos for each assigned person across their task dates, working around what they're already booked on so nobody goes over capacity.
Keep it realistic: no more than [3-4] hours a day on this job for any one person. Nothing should be scheduled during our [closure name], [start date] to [end date].
When you're done, give me a week-by-week summary of who's working on what, and flag anyone who's over capacity in any week.
🔧 Set up:
Assign people first. Run prompt one, or assign team members to the job's items yourself.
Keep your existing Schedule accurate. Your assistant works around current To Dos, so anything missing from the Schedule won't be accounted for.
Set your own daily cap and closure dates. Adjust the hours to suit how your team works, and remove the closure line if you don't need it.
Expect a check-in. Your assistant may pause to confirm before it books a batch of To Dos. Approve it and it carries on.
🔑 Permissions:
View Jobs: to see the job and the Schedule.
View Team To Do and Edit Team To Do: to see what people are already booked on, and create To Dos for them.
Prompt three: the client moves the goalposts
⏱️ Timestamp: 15:47
One date moves and everything after it has to move too. This prompt shifts an item and everything that follows it, moves the milestones and reschedules everyone's To Dos to match, keeps your closure clear, and tells you the new final sign-off date and whether anyone is now overbooked. One change, a lot of knock-on work, handled in a single prompt.
Prompt:
The client has pushed back the [item name] on [job name] by [a week]. Move '[item name]' to start [one week] later, and shift everything that comes after it by [a week] too: [the phases that follow]. Keep our [closure name], [start date] to [end date], clear.
Move the milestones and reschedule everyone's to-dos to match. Then tell me the new final sign-off date, and whether the change leaves anyone overbooked.
🔧 Set up:
Use the item's exact name, as it appears on the job.
Name the phases that should move. Being specific stops earlier work from shifting by mistake.
Add a final sign-off milestone to the job if you want the new date reported back.
Watch out for any company-wide closures. A short slip just before a holiday break can push a whole phase past it, so check the new dates in the Schedule, or add it to the prompt.
🔑 Permissions:
Edit Jobs: to move the items, phases and milestones.
View Team To Do and Edit Team To Do: to reschedule everyone's To Dos and check for overbooking.
Prompt four: cover unexpected leave
⏱️ Timestamp: 19:21
Someone needs time off at short notice. Two things need to happen: their work goes to someone else, and their week is blocked so nothing else gets booked in. This prompt does both, and rebalances the job's planned hours so the plan reflects who's actually doing the work now.
Prompt:
[Team member] needs to take the week of [date] off unexpectedly. Find their to-dos on the [job name] job that week and move them to someone else in the same role who has capacity. Only use actual team members, not placeholder or freelancer users.
Rebalance the planned hours on the job so they reflect who's actually doing the work now. Block out [team member]'s week too with personal to do's for 'leave', so they don't show as available for anything else.
🔧 Set up:
Make sure someone else shares the role. If nobody else has that role, tell your assistant who should pick up the work instead.
Want to cover more than one job? Remove "on the [job name] job" to move all of their work that week.
🔑 Permissions:
Edit Jobs: to rebalance the planned hours on the job.
View Team To Do and Edit Team To Do: to move their To Dos to someone else, and add personal leave To Dos to their week.
Prompt five: check capacity for a new job
⏱️ Timestamp: 22:00
A new job is on the table. Can you take it on? This prompt uses an existing job's roles and hours as a template, checks availability by role across the new job's dates, and sorts the answer into roles that are fine, roles that are tight, and what you'd need to move or bring in. A confident yes or no, based on your actual schedule.
Prompt:
A new client wants a [type of] job similar in scope to [reference job name], but around [half] the hours. It would start on [start date] and run for about [six weeks]. Based on the roles and hours on the [reference job name] job, do we have capacity for it? Check availability by role across that window, including what's already scheduled and in the pipeline. Tell me which roles are fine, which are tight, and what we'd need to move or bring in to make it work.
🔧 Set up:
Pick a reference job planned by role, with hours. That's what the estimate is built from.
Keep upcoming jobs in Streamtime with dates and planned hours. Work that isn't in Streamtime can't be counted in the pipeline.
This one only reads your data. Nothing changes in your account, so it's a safe one to try first.
🔑 Permissions:
View Jobs: to read the reference job and the jobs in your pipeline.
View Team To Do: to see what everyone's already booked on.
No edit permissions needed, as nothing changes.
Related Articles