I'm Anthony, an AI automation specialist and full stack developer in Port Harcourt. I build private AI systems and automations for the bookings, follow-ups, payments and reports your team still chases every day.
WAT (UTC+1). Within an hour of the UK, Germany, France and Spain all year.
This is an AI assistant and can make mistakes. To confirm details, email anthony@stellaxafe.com.
An introduction, and a walkthrough of the StellaXafe Platform.
Products real businesses use. I build them with AI coding agents, and I direct the work and check every result.
A guest experience and CRM tool for hotels. An unhappy guest scans a QR code and reports the problem privately while still on site. The manager gets a WhatsApp alert straight away and can fix it before it turns into a one-star review. It also keeps guest records, takes Paystack payments and pays partners automatically each month. I built it alone in under three months.
A private AI for one company at a time. Five agents work in sequence: they interview staff, read their understanding back to them, then build the Word, PDF, Slides or Excel file. A WhatsApp support agent answers from the company's own knowledge base and passes anything it can't answer to a person.
A trust passport for the digital market in Africa. A seller verifies once and gets a record that grows with each closed deal and honest review. Buyers can check it before they pay.
Open XALOMHears a Bible reference the moment a preacher says it and puts the verse on screen. Runs offline with nobody operating it.
The chat on this page. A visitor's question goes to a Make scenario, which sends it to Gemini with a fact sheet about my work and strict rules, then returns the answer. It stays on topic and ignores instructions hidden in visitor messages.
Most of the work happens before any code is written.
Rule files make each AI agent click through the flows and screenshot the interface across screen sizes before I review anything.
When something already works, I have the AI add only the new lines. If it breaks, the cause is easy to find and I can revert.
For anything that costs money or must happen once, like a message or a payout, I set a database rule. A retry then can't send or spend twice.
Each project keeps a README plus CURRENT, SESSIONS and PENDING-DEPLOYS files, so a new engineer or agent can pick up in minutes.
I keep a production branch and a testing branch, save known good states, and revert first. Then I work out what went wrong.
When a platform is new to me, I screenshot the confusing screens and share them with the AI. Three or four are usually enough to work on my own.
Get in touch if any of these sound like you.
You're building with AI and need scoped, private AI systems
You want to learn how to use AI tools effectively
You're interested in building safety guardrails for AI
You want to build real products, not just use AI
You like solving real problems and working with good people
Tell me what your team repeats by hand and I'll tell you what can run on its own.
