AUTOMATION · AI ASSISTANTS · INTERNAL SYSTEMS
I build small businesses the systems and automations that replace the repetitive manual work, the copy-paste between systems, and the "hold on, let me check and get back to you".
30 MINUTES · NO SLIDES · NO COMMITMENT
FROM A REAL PROJECT: SCANNING TECH ROLES
RELEVANT ROLES FOUND PER MONTH
WHAT I HAVE BUILT SO FAR
WHO WE ARE
Most AI products on the market were built for organisations with an IT department and an annual budget. I build the same things, at the scale, the price and in the language of a business that runs itself. I start from the process that breaks your day, not from a feature list.
What I found
I read job ads from small Israeli firms and checked what the local practice-management software actually covers. Each of the four fields has one clock running in the background that no system counts for you. The software does the heavy lifting, and rightly so. What stays in people's hands is the small work around it.
Accountants
23 September
Three times a year a firm is measured on the share of its client book it has filed. One client who has not handed over receipts is not only holding up himself, he touches everyone else's quota. And whoever chases him does it from memory.
Lawyers
"Where is my case?"
The call that comes in every day. Not because the client is anxious, but because he has no other way to know. I found no Israeli practice-management system that lets him see for himself.
Insurance agents
7 business days
The entire window in which a pension clearinghouse report stays available before it is deleted. Miss it and you pay again and start over. I found no system that warns about the expiry.
Real estate agents
Two months
An exclusivity agreement lapses on its own after two months without two marketing actions. Every property carries a different date, and an office with dozens of them is tracking hundreds of dates in one person's head.
Each card opens the full pain map, written in Hebrew. All four findings rest on sources that do not move: legislation, arrangements with the tax authority, and officially published information.
SERVICES
Every project starts with a short scope document and a fixed price. No open-ended hours, no surprises on the invoice.
A repetitive manual process someone does every morning, moved to running itself in the background.
A language model wired into your own data and procedures. It answers, summarises and extracts on your behalf.
For when the business is too big to run by hand and too small for enterprise software that costs a fortune.
WHY MANOMATIC
A short scope document with a price and a finish date. No open-ended hours, no "it came out a bit bigger than expected".
You describe how the business works and what annoys you. Translating that into technology is my problem, not yours.
If a simple script solves your problem for a quarter of the price of an AI system, I will build the script and tell you so.
On delivery you get the code, the database and every credential in your name. You are never locked in to me.
CORE SOLUTIONS
Each one is built to fit your actual process. Most go live within two weeks.
Scans your information sources, filters by your rules, and sends only what's relevant.
REPLACES: THE DAILY MANUAL SWEEPAnswers customers on WhatsApp or your site, drawing on your own content and procedures.
REPLACES: ANSWERING THE SAME QUESTIONSInvoices, receipts and orders are read automatically and filed where they belong.
REPLACES: MANUAL DATA ENTRYEvery incoming message is classified by topic and urgency and lands with the right person.
REPLACES: SORTING THE INBOXA new lead gets a reply, the right qualifying questions, and a slot in your calendar.
REPLACES: PHONE TAGThe monthly report builds and sends itself, with the numbers always current.
REPLACES: A LOST DAY EVERY MONTHWhatsApp, email and phone arrive in one place, with the full history for every client.
REPLACES: FOUR OPEN APPSOne screen showing where every client stands and who hasn't been answered in time.
REPLACES: "WAIT, WHAT'S THE STATUS?"WORK
Three systems I built end to end. The descriptions and outcomes are real; the screens are styled illustrations of them, with no real data.
The entire client lifecycle, from closing the deal through the annual work plan to ongoing advisory, ran on spreadsheets and email threads. Managers could not tell you a client's status without asking three people, and stalled clients fell through the cracks.
An end-to-end management system: role-based permissions, annual work plans with manager approval, client-facing task management, actual-hours tracking against agreed expectations, a proper handover process between consultants, and a portal the client logs into.
The main screen does not list clients. It answers one question: who needs attention today. In daily production use.
A manual sweep of several sources every morning. Once automated, a bigger problem surfaced: “nothing today” and “this source has been broken for six weeks” look identical. Four broken scrapers went unnoticed that way.
A service running hourly in the cloud, scanning sources in parallel, filtering against predefined criteria and de-duplicating against everything already sent. On top of it, a daily health report stating what each source can actually produce right now.
The same scan went from 2 relevant results to 46, and a second source from 8 to 43. Same code, same sources. The only change was knowing something was broken.
Revenue and expenses lived in a file updated once a month, if that. “How much came in this month?” took half an hour of reconstruction from the calendar and the receipts.
A web app with categories, recurring costs and a forecast, plus the one thing that made the difference: automatic sync of calendar appointments straight into revenue lines.
Most of the bookkeeping happens without anyone touching it, and the answer takes a second.
FAQ
Entirely dependent on scope. A single automation and a full management system are different worlds. The discovery call gives you an honest range; the scope document gives you a fixed number. I do not start building until you have approved it, and there are no open-ended hours.
A simple test: if someone on your team spends more than an hour a week copying data between files, a basic automation pays for itself inside a year. If it does not, I will tell you that on the call instead of selling you a project.
Most of the time, yes. That is why I use AI only where it genuinely solves something: reading documents and invoices, answering from your own content, summarising and classifying enquiries. If a simple script solves your problem for a quarter of the price, I will build the script.
Entirely. At the end of the project you receive the code, the database and every credential in your own name. You're not locked in. You can continue with any developer.
The first month of support is included, covering fixes and small changes. After that you can move to a monthly retainer, or work on demand at an hourly rate agreed in advance.
The data stays in your accounts, not mine. When I use a language model I work through business APIs that do not train on your content, and I expose only what the task actually requires.
Sometimes you genuinely should, and I will say so. Custom pays off when your process is unusual, when off-the-shelf tools force you to work in a way that doesn't fit the business, or when the monthly subscriptions to several tools already cost more than building it once.
GET IN TOUCH
Tell me which manual process annoys you most. By the end of the call you'll know whether there's something worth building, what it costs and how long it takes.
