Most of the market hasn’t moved yet.
Only one in five EU companies with 10+ staff used AI at all in 2025. The agencies that build it into delivery now get the head start.
SOURCE: EUROSTAT, DEC 2025 ↗Reporting, research, content and client updates eat
the hours you can’t bill. A 4-week sprint automates
them on your real client work, measures the time
saved, and trains your team to run it.
Built and taught personally by Alexander, an engineer, not a prompt course.
Illustrative workflows. Not a live AI tool or a client result.
Not another AI tool.
Not a prompt course.
Automations built on your client work.
An agency sells expert hours. When the routine part of the work gets faster, the agencies that changed their workflow first can take on more clients, turn work around faster, or protect their margin.
Only one in five EU companies with 10+ staff used AI at all in 2025. The agencies that build it into delivery now get the head start.
SOURCE: EUROSTAT, DEC 2025 ↗A few people use AI their own way, results vary by person, and nothing reaches the client process. The time saving stays small and the quality risk stays high.
Every new client usually means more account and production hours. Take the repetitive part out first, and the team you have can carry more of the load.
Not a list of tools to try. Working automations on your real client work, tested against how you do it today, and a team that can run them without me.
Reporting pipelines from your ads and analytics data, research agents, content checks against client style guides. I build them with your team, on the tools you already pay for where possible.
Ready-made workflows for reporting, research, content operations and client updates, with review checklists and rules for client data. Adapted to your clients and tools.
Each experiment is timed against your baseline: hours per report, revision rounds, turnaround, tool cost. You see what to keep and what it’s worth.
One team, real client work, a fixed timeline. By week four your people run the new workflows, the client-facing quality checks are in place, and you know what each one saves.
Map where time goes across accounts. Pick 3–5 repetitive workflows and record how long they take today.
↳ Experiment backlog + baselineBuild an AI workflow for each one with your team, on your tools and client data rules. Run them on live client work.
↳ Measured result per workflowHands-on sessions on the winners. Account managers, analysts and writers practise on their own clients. Name your AI champions.
↳ Trained team + internal championsWrite the winners into your agency handbook, set the quality checks, and plan the next 90 days of experiments.
↳ Handbook + rollout planTime saved is measured on your client work, not promised on a slide.
Some experiments will lose. We drop those and say so.
Most agencies try one of these first. Here’s what usually happens.
More salary and onboarding time, and the same manual process, just with more people doing it.
Take the repetitive work out first. Decide on the hire once you know what’s left.
Another subscription that a few people try, which doesn’t fit how your agency delivers.
Workflows built around your clients, templates and tools, using what you already pay for where possible.
Generic demos. Nothing changes in Monday’s client work.
Your accounts, your reports, your style guides. The work changes in the same week.
Results depend on who did the work, and nobody checks what goes to clients.
One shared handbook, quality checks before anything ships, and champions to keep it going.
15+ years in production engineering. Now I help teams work AI-first.
I’ve led engineering teams, built AI agents and automation, and created a 16-week programme that trained more than 20 engineers. That means the automations I build for your agency are built to hold up, not to demo well.
I sit with your team, build the workflows, check the output, and teach people to run them. Every session is delivered by me.
Take the routine work out.
Keep the quality in.
Digital marketing, performance, SEO, content and web agencies, roughly 10–200 people, with recurring work for several clients. If most of your hours go into reporting, research, content production or client coordination, there’s usually something worth testing.
Not if it’s set up properly. The workflows run from your own client briefs, style guides and past work, and a person reviews everything before it reaches a client. The aim is to take out the repetitive part so your people spend more time on strategy and creative judgement.
The first workflows run on live client work in week two, and you get before-and-after results by week four. I don’t promise a percentage in advance. We measure it on your work, including review and revision time.
Plenty of places. Strategy, client relationships and creative calls stay with people. AI output can be wrong in confident ways, so every workflow keeps a human check. Some experiments will lose to the current way of working; we drop those and say so in the results.
We agree data rules before anything touches client data: which tools are approved, what can be shared, and what stays out. Early examples can use public or sanitised material. Your contracts with clients set the limits, and the handbook writes them down for the team.
Usually not much. We start with the tools you already pay for. If an experiment needs something new, you see the cost next to the time it saves before you decide.
When you book a call, Cal.com handles the booking and receives the details you enter there. This site only records that a call was booked and which campaign link was used, with no name or email. No mailing list, no tracking or analytics, and no cookies are set by this site. The Cal.com calendar, which may set its own cookies, loads only after you accept in the cookie notice or click “Load calendar”. To have your booking details removed, ask Alexander on the call or reply to the calendar invite.
Fees depend on team size and scope. A written proposal sets out the fee, sessions, deliverables, tool costs and responsibilities before you commit.
Book a 20-minute discovery. We’ll look at where your team’s hours go across clients and which workflows AI could take on first.
Where your team’s hours go across clients today.
The 2–3 delivery workflows AI could take on first.
Whether a 4-week sprint makes sense. If it doesn’t, I’ll say so.
You leave with a shortlist of workflows worth automating, whether or not we work together.
Pick a timeOpens the calendar. Booking is handled by Cal.com. Privacy