Modeled scenario · Marketing agency · 14 employees
A 14-person marketing agency
target · 28 hrs/week handed off
Two AI employees draft the client-status and project-ops answers, and two account managers get their week back.
The starting point
The agency runs twelve active client projects at a time. Every day account managers answer the same three questions: where is the deliverable, can we move the review, has the invoice been paid. In this scenario those interruptions add up to roughly six hours a day across the team, and the two account managers who should be selling are trapped in coordination.
The deployment
The agency scopes two AI employees against the two loops that eat the day: client status and project ops. Each answer is drafted with the current project state attached, and the account manager sends it. Anything that changes scope, price or a deadline goes to a person before it goes to the client.
- Status answers drafted from the current state of the project board.
- Reschedule requests drafted against the agency calendar, with the conflict named.
- Invoice questions drafted from the payment record, with the amount marked pending until a person confirms it.
- Scope changes and sensitive threads escalate to the account manager untouched.
The targets we model
In this scenario the assistants cover most routine client communication within the first month, and account managers work the queue twice a day instead of reacting all day. The tiles below are the targets a pilot of this shape runs against.
28
Team hours handed off per week
70%
Routine client questions answered from a draft
Down 60%
Account manager reaction time
2
Account managers freed for new business
Targets for a pilot of this shape · not measured results
Coordination is the automatable layer. The client relationship stays with the account manager; the AI keeps the status noise off their desk.
Why this scope works
Creative work stays human. One source of truth for status questions removes the noise that hides the real work, and the freed hours go to client strategy and pitch calls.
Method · AI support + ops
How this scenario is built
Nothing above came from a customer account. It came from four inputs and one multiplication, both shown here, so you can check the arithmetic and then argue with the assumptions.
01 · Input
The routine loop we start from
The “Customer support” row of the ROI calculator puts 40 hours a week into the repeatable part of this job. The scenario runs two employees against that loop.
02 · Input
The share we assume moves to the machine
The share of that loop the AI takes on. It is the one number we choose per scenario, and it is the slider you move on the calculator — everything else follows from it.
03 · Arithmetic
Hours freed, and what they cost today
At $35 an hour loaded — salary, taxes, tooling and management time — 28 hours a week is $4,243 a month of routine time, against $298 a month for two employees. Same formula as the calculator, same weeks-per-month constant of 4.33.
04 · Targets
The percentages are targets, not measurements
Uplift in meetings, resolution share, no-show rate: each one is a target agreed before a pilot starts and then measured in your own systems. No figure on this page comes from a customer account.
05 · Assumed
What this scenario assumes we would connect
The scenario assumes a connection to the agency's project board and shared client inbox. Unistaff ships neither connector today, so a deployment of this shape starts by scoping them.
Same inputs, same formula — swap in your role, your hours and your loaded rate.