I never planned to build a business around AI. I started where many operations freelancers do: cleaning up chaotic processes for clients, building spreadsheets, documenting workflows, and helping them make sense of messy operations. AI arrived as a curiosity. Then, almost overnight, it became the most powerful lever in my work.
Today, I lead EKB Labs, an AI consulting studio that assists mid-sized businesses in transforming their processes through automation, agents, and improved information flows. The focus is very simple: cut noise, reduce manual work, and give teams tools that actually get used in the real world.
My path into AI began with one small experiment. A client in the real estate industry needed to turn dense training material into practical exam-style questions and checklists for their agents. I started using a language model to draft questions, refine explanations, and restructure content into drive time learning. It saved hours. More importantly, it made their expertise easier to reuse and scale.
From there, I leaned in fully. I shifted my Fiverr profile toward AI strategy, automation, and training, and built EKB Labs to support more in-depth projects.
On Fiverr, I now sit in an interesting middle ground. I still do one-on-one consulting sessions, but many of my clients are established businesses with complex systems. Examples include law firms that want AI dashboards to surface case insights, training companies that need content engines rather than single prompts, food service businesses that need orders, staff rotas, and stock to seamlessly communicate with each other. In each of these instances, I’ve helped them zoom out, map their workflows, then design AI that supports those flows rather than fights them.
Over time I have built an 80% client retention rate, with some clients returning for 15+ projects, because the systems I deliver keep working and evolving with their businesses.
In practice, that means a mix of tools. For conversational work, I use the major models inside structured prompt systems, often with retrieval over the client's own documents. For operations, I lean on tools like n8n or Make to connect CRMs, forms, voice agents, and databases. For data, I like simple but robust stacks such as Postgres or vector stores, sometimes sitting behind a lightweight API. Around that, I add what most people overlook. A clear front door for users, a way to capture feedback, and simple analytics on how the AI is actually used.
Some of my favourite builds have been voice agents that sit on top of this stack. For example, a catering client now has a voice assistant that can answer questions about their menus, pricing, and availability, and can also log leads directly into their CRM. Under the hood, it’s just good call routing, a retrieval layer, and a clear decision tree. To the user, it feels like magic. To the business, it feels like new capacity that does not burn out.
Another voice system I deployed for a mid-sized EU company cut their customer service response time by 88%, from 22 minutes down to 2.5 minutes, while adding 24/7 coverage. That system now saves them €1,900 per month in operational costs and hit a 4.2x return on investment within the first 90 days.
The biggest lessons have been around behaviour, not models. Clients do not need another shiny chatbot. They need a reliable colleague who does what it says on the tin. That means tight scoping, honest conversations about what AI can and cannot do, and designing for failure paths. I spend a lot of time helping teams write better prompts, set guardrails, and build small experiments before big rollouts. Training and change management often matter more than the model choice.
Balancing creativity, strategy, and AI has become the core of my work. I use AI heavily in my own business for research, outlining, and drafting content. I treat it as a junior partner that can move fast and suggest options. My job is to set the direction, decide the constraints, and do the final edit. The same applies to client systems. AI can draft emails, propose next actions, or suggest automation rules. The human decides what is right for the context.
For other freelancers curious about AI, my advice is to start by identifying your own bottlenecks. Automate the boring parts of your delivery first. Turn your repeated answers into prompt templates. Build a small knowledge base of your own work and let a model search it. When you feel the benefits yourself, it becomes much easier to sell that value to clients. Pick one service where AI gives a clear outcome, such as faster research, better proposals, or simple process automation, and package it clearly on your gig.
The market is noisy, so positioning matters. I have found that clients respond better to outcomes than buzzwords. I rarely sell "AI". I sell fewer manual hours, faster response times, clearer reporting, or new revenue streams. On Fiverr, this might appear as a consultation offer, a fixed-scope automation build, or a training session for a client's team. Once trust is built, larger strategy work often follows.
Looking ahead, I see EKB Labs becoming less about one-off projects and more about repeatable systems. I am building internal playbooks, reusable components, and my own AI assistant that understands my processes and past projects. The goal is simple. Spend less time reinventing the wheel and more time on the hard, valuable parts of client work.
As AI models evolve, freelancers who understand business workflows, communication, and real constraints will stand out. That is the space I plan to stay in and grow.
