Client reporting
GA4, Google Ads and Meta data pulled every week, an AI-written summary, a branded report ready for a 5-minute check.
AI automation for agencies
I give agency teams back 5+ hours per person, every week.
Custom AI automation for the work your team repeats: reports, lead intake, onboarding, content drafts. Built in real code, running on your own accounts.
- Top Rated on Upwork · 100% Job Success
- 10+ years as a developer
- Founder of Practikk, a SaaS run on automation
The math
What is manual work costing you?
Move the numbers to match your business. Nothing is saved or sent anywhere.
Your hour is worth
$0
monthly profit ÷ hours you work
Repetitive work costs you
$0
Automation could save you
$0
Repetitive work costs your team
$0
Automation could save your team
$0
An estimate: only the share of each task that can typically be automated is counted.
What I automate
The weekly work nobody should do by hand.
Lead intake
Every inquiry lands in your CRM in seconds, summarized and scored by AI, with the right person notified.
Client onboarding
A signed proposal sets up folders, channels, the project and the welcome email. Nobody follows a checklist.
Content workflows
AI drafts briefs, captions and recurring formats. A human approves before anything goes out.
Proposals and follow-ups
Drafts from call notes, reminders that never forget, a pipeline that updates itself.
Your agency on one screen
Hours, clients and profitability in one dashboard, instead of seven tools and a gut feeling.
I use Make or n8n when they fit, and write real code (Python, Node.js) when they don't. You get the right tool, not the only one I know.
Examples
What it looks like when it runs.
Two sample systems I built on demo data, so you can see the result your team would get. In both, AI writes the drafts and you decide what gets sent.
Sample project · demo data
No more chasing clients for logos, copy and logins
- Each project gets a list of what you need from the client, with due dates.
- When something is missing, AI drafts a clear, friendly email to the client.
- You read it and press send. Nothing goes out without you.
- It keeps track and drafts the next reminder until each item arrives.
Goal: projects stop waiting on clients, and nobody writes the same reminder five times.
Waiting on clientUpdated today
Little Oven Bakery · new website
- Logo filesReceived
- Homepage copy2nd reminder drafted
- Domain loginDue Thursday
- Shop photosLate · call suggested
Draft for Maya: Hi Maya, quick nudge on the homepage copy. Once we have it, we can show you the first full version next week. A rough draft is perfectly fine.
SendEdit
AI drafts the reminders, you approve before anything is sent · demo data
Sample project · demo data
Overdue invoices followed up for you
- Every morning it checks your invoices in Xero, QuickBooks or Stripe.
- A few days late: AI drafts a friendly reminder with the payment link.
- Still unpaid: it drafts a firmer second one, in your tone of voice.
- You approve each draft with one click. At 30 days it tells you to call.
Goal: paid faster, without writing awkward emails.
Unpaid invoicesChecked 8:00 AM
$9,850 outstanding
| Client | Amount | Status |
|---|---|---|
| Northfield Coffee | $2,400 | 3 days late · draft ready |
| Ortiz Physio | $4,200 | 14 days late · 2nd draft ready |
| Ledgerly | $1,800 | Paid after reminder |
| Little Oven Bakery | $3,250 | 31 days · call them |
Review 2 drafts
AI drafts the reminders, you approve before anything is sent · demo data
Proof · my own business
I automated my own SaaS first.
Practikk is a subscription product I built and run: monthly business content plus an AI agent that applies it to each subscriber's business. Here is what changed when I rebuilt it and automated the work of running it.
| Before | Now | |
|---|---|---|
| Stack | Webflow (2 plans), Xano, Stripe, Omnisend, Cloudflare proxy, separate landing site | One Next.js app with Supabase and Stripe |
| Tools per month | about $200 | about $40 |
| Monthly content | about 20 hours | about 5 hours. AI drafts, I review |
| Support replies | written from scratch | AI suggests a reply with customer context. About 30 hours a month saved |
- Daily delivery of content to subscribers
- Subscriptions, credits and failed payments
- Cancellation flow with a retention offer
- Email marketing events and conversion tracking
- Admin dashboard with the numbers I check every morning
The rule I use everywhere: AI drafts, a human decides. Nothing goes to a customer without a person on the send button.