Connected to the real tools
Outlook, Xero, Dropbox and Slack. It finishes the work in the systems the practice already uses, instead of explaining how to do it.
Case study: AI agent install
Getting the admin off a designer's desk
How a regional NSW landscape architecture practice uses Viktor, an AI co-worker in Slack, to handle the inbox, invoicing, fee proposals and tenders, so the principal's time goes back into design.
2 promptsto a branded PDF fee proposal with project photos
- 96pieces of work handled in one month
- 62inbox briefings, none missed
- 7routines running on their own
- 0emails or invoices sent without approval
- Drafts only: a person approves everything that goes out
- Works in Slack, Outlook, Xero and Dropbox
- Landscape architecture practice, regional NSW
The standout
A fee proposal in a couple of prompts
A polished fee proposal with images is work that would take hours in expensive design software. Now a couple of prompts produce a professional-looking PDF, branded exactly like the practice's own.
Viktor
Fee proposal PDFbranded, with project photos
- The practice's own format
- Past project photos
- The principal sets every fee
Spoken from the car between site visits. Waiting back at the office as a draft.
A couple of prompts, one polished proposal. The principal gives Viktor the client, the site and the hourly rate. Back comes a professional-looking PDF, branded exactly like the practice's own, with photos of past rural and acreage projects already laid out.
One more prompt for the numbers. Need a cost breakdown on the letterhead to go with it? Viktor prepares it as an editable Word document.
The principal stays in charge. Every fee is set by the principal and every send is approved. Viktor just does the assembling.
One month
What one month looked like
These figures come from the assistant's own monthly report. It says plainly they're a close approximation, not an exact count. It doesn't round itself up.
96
pieces of work handled in a single month
What gets asked for
- Fee proposals
Professional, branded PDFs with past project photos.
- Invoices in Xero
Maintenance invoices prepared as drafts to check.
- Documents on the letterhead
Cost breakdowns, project summaries and a species list for a report appendix.
- Finding old records
Like an engineering drawing from 2017, buried somewhere in Dropbox.
- Calendar and spreadsheets
Dates added to the calendar and an editable site visit spreadsheet.
On autopilot
What runs on its own
Seven routines run without anyone deciding to do them. Most of the value is in things just happening.
| Routine | How often | What it does |
|---|---|---|
| Every day | ||
Inbox briefing | 2xdaily | Reads the office inbox, summarises what's new, flags what needs the principal and matches senders to existing clients in Xero. Files obvious marketing mail out of the way and lands as a message in Slack, morning and afternoon. |
| Every week | ||
Maintenance invoicing | Autowhen hours land | Prepares the maintenance invoice as a draft in Xero, ready to check. If something is missing, like the rate, it asks instead of guessing. |
Maintenance hours reminder | Wedweekly | A short, friendly nudge in the maintenance channel asking for the week's hours. |
Council tender scan | Monweekly | Searches open NSW council landscape architecture and design tenders. Marks the new ones and lists closing dates, local councils first. |
Client list refresh | Friweekly | Keeps client and customer records current from Xero and flags anyone new. |
Usage check-in | Thuweekly | Scores how well the assistant is being used and suggests ways to get more out of the monthly allowance. |
| Every month | ||
Monthly report | 1stmonthly | A plain-English summary of what it's set up to do and what it actually did, emailed to the office. |
Drafts only.
Nothing is sent from the inbox or authorised in Xero without approval. The one exception is a choice the practice made: filing marketing mail now happens without asking.
Two moments
The bits that stood out
Standout
From one request to a weekly routine
Asked once
The principal asked for a search of open NSW council tenders.
Useful enough to keep
The results were good enough to make it a standing routine.
Every Monday morning
New tenders marked, closing dates listed, local councils first. Nobody has to remember to look.
Trust
It asks rather than guesses
Hours come in
Twice in the month, maintenance hours arrived without everything needed to invoice them.
Something is missing
Instead of filling in a number and hoping, it noticed the gap.
It asks
Both times it messaged to confirm the hours or the rate. That's what you want from anything touching your invoices.
Why it works
Why it works in a small practice
This isn't replacing anyone in the practice. It's taking the admin layer off the person whose time is best spent designing.
People approve, it drafts
Emails and invoices stay drafts until someone says yes.
Scheduled, not remembered
Daily, weekly and monthly routines run without anyone deciding to do them.
It asks rather than guesses
When a rate or a number is missing, it checks with a person first.
It reports on itself
A weekly check-in and a plain-English monthly report on what's running and what it actually did.
It only sees what it's invited into
Project channels stay private until the practice chooses to add the assistant.
What's next
Ready when the practice is
These skills are set up but not switched on yet. Each one is a single message away.
Drawing pre-checks
A checklist of missing initials, title block issues and sheet numbering gaps before a set goes to the client or council.
Ready to switch onGoogle review replies
Unanswered reviews found, each with a short reply drafted for approval.
Ready to switch onSite meeting notes
A recording turned into notes, decisions, an action list and a draft follow-up email.
Ready to switch onCase studies
More case studies
The admin was burning out the volunteers
How the Southern Highlands Chamber of Commerce uses an AI co-worker to run events, member communications and reporting, so volunteers get their evenings back.
Read case studyThe reminder customers were actually waiting for
Sales improved by 320% and orders by 626%. Orders grew faster than revenue because the reminders brought back a large number of existing customers for a repeat purchase, which is exactly what the campaign was designed to do.
Read case studyCurious what this would look like in your business?
Book a free discovery session. We'll show you exactly how Viktor would work in your business, and tell you honestly if it's the right fit. No commitment.