- Table of contents
- 1. The trigger: why I wanted to automate everything
- 2. What actually worked
- 3. My costly mistakes
- 4. The necessary rebalancing
- 5. Where the agency stands today
- FAQ
Lionel Anderson
Founder, Mister Anderson — AI Agency in the Basque Country
12 months ago, I made a decision that changed my agency: automate everything with artificial intelligence.
Between the euphoria of early wins and the painful disillusionment, here's my unfiltered story.
Table of contents
1. The trigger: why I wanted to automate everything
It was June 2025. I'd been running Mister Anderson for a few months, and I was suffocating. Quotes,
client follow-ups, content production, reports… I spent my weekends catching up on
what I hadn't done during the week. I felt like I was running after my own business
without ever catching it.
By stopping to think for a moment. I told myself: if AI can write an email in 10 seconds, why am I still
losing 20 minutes on every follow-up? If it can generate a report, why am I
losing half a day every Friday compiling data?
The idea hit me like an obvious truth: turn Mister Anderson into an AI agency
fully automated. A well-oiled machine where humans would only be there to supervise.
I wanted to work 4 hours a week, like Tim Ferriss. I wanted to scale without hiring.
I finally wanted to breathe.
My promise to myself: in 6 months, 80% of operational tasks would be
automated. I wasn't holding back.
2. What actually worked
I won't lie to you: some AI agency automations were
game-changers. Here are the ones that transformed my day-to-day, and that still work today.
First-draft content generation
Before, writing a blog post used to take me an entire day. Today, I produce
80% of the initial content with AI in under an hour. It's not perfect — I'll come back to that —
but it gives me a solid base to work from. I've quadrupled my content
output without sacrificing final quality.
Here's my current workflow:
- I give the AI a detailed brief with my keywords and my angle
- It generates a structure and a first draft
- I rewrite everything entirely in my own voice, keeping the good ideas
- I fact-check and enrich it with my personal experience
Automating client reports
This used to be my nemesis. Every Friday, I spent three hours pulling data from Google
Analytics, formatting it in PowerPoint, writing up comments. Today,
a Python script connected to the API generates a complete report for me in 10 minutes. All I have
to do is add my strategic recommendations.
Real time saved: 3 hours → 20 minutes per report. Across 15 clients,
that's 37 hours saved every month.
Scheduling and calendar management
I connected Calendly to my CRM, which automatically syncs my availability.
Prospects can book directly, receive a confirmation email personalized
by AI, then a reminder 24 hours before. I no longer have to manage a single email exchange
to set up a meeting.
Answering frequently asked questions
I trained an AI assistant on my knowledge base. It now automatically
answers recurring questions: "What are your rates?", "How long
does a project take?", "Do you work with small businesses?". This filters out 60% of
incoming requests, and I only step in for qualified conversations.
3. My costly mistakes
I'm not telling you all this to show off. Because behind
these wins, there are failures. Big failures. Mistakes that cost me clients,
money, and a lot of sleepless nights.
Over-automating the relationship
My first big crash happened in September 2023. I'd automated my quote
follow-ups with an AI email sequence. The result? A potential 30,000-euro
prospect sent me packing. His email said: "Lionel, this is the third time you've sent me
the same robotic message. I don't want to work with a machine."
I'd wanted to scale so badly that I'd forgotten the essential: the human relationship.
People don't pay good money to interact with an algorithm. They pay for my
expertise, my attentiveness, my ability to understand their specific problems. By
automating the relationship, I had diluted my value.
Lesson #1: AI agency automation should never touch the moments
of truth with your clients. Prospecting, negotiation, strategic follow-up:
those stay human, or they don't work.
Losing the human touch in content
For three months, I published articles that were almost entirely AI-generated.
My traffic went up. My engagement dropped. Comments became rare.
Shares, non-existent. I had volume, but no soul left.
One day, a loyal reader wrote to me: "I no longer recognize your voice in your
recent articles. It's become generic." He was right. I had sacrificed my
authenticity on the altar of productivity.
AI's costly mistakes
AI hallucinates. It makes up statistics. It suggests strategies that don't fit.
Once, it recommended that a real estate client target "students
under 25" for luxury property purchases. Thankfully, I reviewed it before sending.
But imagine the credibility lost if that report had gone out?
I also discovered that AI tends to suggest the same solutions to everyone.
It lacks context, nuance, deep understanding of business challenges.
It's an excellent assistant, but a very poor strategist.
4. The necessary rebalancing
In early 2024, I took a break. I stopped all my automations for
two weeks to start from scratch. I asked myself a simple question: what actually
creates value for my clients?
The answer was clear: tailored strategy, a deep understanding of their
business, creativity in solving problems, a relationship of trust.
Everything else — formatting, data collection, repetitive tasks —
could be automated.
So I redesigned my model. Today, AI handles 70% of operational execution,
but 0% of strategic thinking and client relationships. I spend less time
on tedious tasks, and more time on what really matters.
My golden rule now:
- Automate: data collection, formatting, repetitive tasks with no added value
- Hybridize: content production (AI for the structure, humans for the soul)
- Protect: strategy, creativity, client relationships
5. Where the agency stands today
Mister Anderson now has ten regular clients, compared to three 12 months ago.
My revenue has grown by 40%. But above all, I work less.
Really less. I leave at 6pm. I take full weekends off. I have time for sports,
something I'd abandoned for years.
AI agency automation isn't a magic wand. It's a
powerful lever, but a dangerous one if misused. It's saved me an enormous amount of time
on operational tasks. It's also made me lose opportunities when
I let it encroach on territory that wasn't its own.
If I had to give one piece of advice to an entrepreneur who wants to automate their agency:
start with what doesn't directly touch your clients. Automate internally.
Keep the human touch where it makes a difference. And above all, always check
what AI produces. It's brilliant, but blind.
AI won't replace agencies. It will transform those that know how to use it
intelligently. And it will eliminate those that believe they can delegate everything to a machine.
To learn more, discover how my AI agency
supports businesses across the Basque Country in their digital transformation.
FAQ
Which AI tools do you recommend for an agency?
For content generation, I mainly use Claude and ChatGPT.
For workflow automations, Make and Zapier are essential.
For automated reports, a Python script with the Google Analytics API
does the job. Finally, for the FAQ assistant, I built an
in-house solution with the OpenAI API, but tools like Botpress work
very well to get started.
How long does it take to set up these automations?
Allow 2 to 3 months for a first effective wave of automation.
The longest part isn't technical — it's the thinking about
what to automate and what to keep human. My advice:
start small. One automation done well beats ten
sloppy ones.
Won't automation dehumanize my agency?
That's exactly the risk I ran, and that I describe for you in
this article. Automation dehumanizes if you apply it in the wrong
places. But if you confine it to operational tasks, it
frees up time for more humanity where it counts. It's a question
of balance, not surrender.
What budget should I plan for automating an agency?
The costs are mostly in internal time. For licenses, budget between
100 and 300 euros a month for a complete stack (AI, automation,
CRM). The real investment is human: someone needs to understand
these tools, configure them, maintain them. If you're not
technical, budget for a consultant or training.
Can you really trust AI for client-facing content?
No, not blindly. AI generates plausible content, not necessarily
accurate. It makes up statistics, suggests unsuitable ideas,
lacks business context. My current workflow: AI generates a first
draft that I rewrite entirely. It's an accelerator, not a replacement.
Never send AI-generated content to a client without thorough human review.
AI agency automation
AI agency
digital transformation
Basque Country
productivity