The most important question isn't:
“What can AI do?”
The better question is:
That is where a good, profitable project almost always starts. Every company has processes that repeat daily: someone sorts the inbox, someone retypes data and checks invoices, someone chases clients about payments, someone makes sure no quote or complaint is left unanswered.
An AI agent is not meant to replace people or act blindly. Its job is to take over the boring, repetitive traffic engineering — always on your terms and under your team's full control (Human-in-the-Loop).
Ex-Google · For small and medium businessesClients come back regularly, but book too late — by the time they call, the calendar is already full.
Read the caseA simple question like “how was traffic last week?” means logging into GA4 and hunting for reports.
Read the caseHappy clients stay silent while unhappy ones post publicly — before you get a chance to react.
Read the caseKnow-how stays in the founder's head, so the team keeps asking the same questions and onboarding drags on.
Read the caseInvoices get lost in inboxes, data is retyped by hand and payment deadlines slip out of control.
Read the caseHours, business travel and per diems settled by hand eat a whole week of HR time and cause payroll errors.
Read the caseOne missed permit expiry date can halt a construction site and trigger heavy penalties.
Read the caseDon't see your process here?
Tell me about it.
What it does: Reads incoming email, matches it against your knowledge base, checks calendar availability or CRM status.
Safety: It sorts and tags messages on its own, and prepares ready-to-send drafts that wait for an employee's approval. Nothing goes out unchecked.
What it does: Picks up invoices from email, extracts data with OCR, verifies the supplier, pre-categorises the expense and prepares the document for your accounting system.
Safety: It catches duplicates and missing data, but final payment approval and sending anything to your accountant always needs a human click.
What it does: Takes the ticket, verifies the order number in your shop, and asks the customer for missing photos or details following your procedure template.
Safety: It builds a complete internal ticket, saving your team from hunting for information across three different places.
What it does: After a quote goes out it keeps the contact schedule, reminds the salesperson to call, and makes sure no lead gets lost in the crowd.
Safety: It works as a digital memory assistant — surfacing notes and dates, while the relationship and the conversation stay with your team.
When we deploy AI-based solutions we design hard operational boundaries from day one:
The agent does not take company knowledge “from the internet”. It answers and acts strictly according to the procedures, policies and price lists you provide. If something isn't in the documents, it doesn't guess.
Critical actions (sending an email to a client, changing a financial status, approving a payment) always require human authorisation. The system prepares a recommendation; a human makes the decision.
The agent can access only the tools and folders needed for one specific task — not a byte more.
Every step the system takes leaves a readable trail, so you always know why the agent suggested what it did.
The process is transparent, safe and pressure-free. Each stage proves its value before we move on.
Together we look at where time leaks fastest in your company and where repetition is highest.
We check whether your data is organised and whether the process is a good fit for AI (if plain, simple automation without AI is enough, I'll tell you straight away).
We start with one concrete slice of work (invoice handling or quote emails, for example) instead of turning the whole company upside down.
We connect the tools, load your rules and test the system on real, historical cases from your business.
We go live with approval gates switched on for your team. We monitor results, measure hours saved and fine-tune the details.
For 15 years I worked at Google, building technology at scale and learning how complex systems can be made reliable, simple and genuinely useful. But some of my most important lessons about technology came later — from working with small and medium businesses, and from becoming an operator myself.
I left Google to run a wellness retreat in Bali.
Suddenly technology wasn't global-scale infrastructure any more. It was about whether a small team could get through the day without drowning in admin. I was the person fixing the boiler at 5am, building the rota, solving operational problems and making sure guests never saw the chaos backstage.
Living on both sides changed how I see technology. I've seen what sophisticated systems make possible — and what running a real business with limited time, people and resources actually feels like.
Today I connect those two worlds. I help companies turn operational complexity into technological clarity, designing practical AI systems that take work away instead of adding it.
Technology should be a quiet, competent partner — not one more tool that needs looking after.
You didn't start a business to answer the same email for the thousandth time. You didn't take the risk to micromanage a spreadsheet at 11pm on a Sunday.
Somewhere along the way you stopped being the head of operations and became the bottleneck. The dream of your own company turned into a treadmill. My goal is to change that — to give owners of small and medium businesses time for creative work, for family, for the deeper “why” that started all of this.
I'm not here to replace the human element of your business. I'm here to set it free.
“The first jump is always the biggest one.”
If you're still reading, something in you already knows it's time. An audit is a way to make that first jump — together, and gently.
You don't need to know how to build an agent technically. Just tell me about the process that frustrates you most in your business. Let's have a short call and find the one small deployment worth starting with.
I usually reply within a day · Limited number of projects per month