When Mark Kashef finished his master’s in management and AI, one lesson stood out above the rest: how to talk to machines. Long before “prompt engineering” became a buzzword, he realized that the real power of AI would belong to people who could ask it the right questions.

A few days after ChatGPT appeared, Kashef turned that insight into a business. He launched Prompt Advisers, an AI automation agency built on a simple promise: teach people how to prompt, not just build tools for them. To test the market, he quietly posted a Fiverr gig offering to “write your prompts for you,” charging just $10 per prompt.

The response was immediate. As early models like GPT‑3.5 Turbo demanded painstakingly detailed instructions, demand for expert prompts exploded. Kashef’s listing climbed to the top of Fiverr search results for “prompt engineering,” pulling in tens of thousands of impressions a month. Prices rose from $10 to $50 to hundreds of dollars for carefully crafted prompt playbooks, and Fiverr’s own PR machine amplified his profile further.

All of this happened while he was still managing a data science team full-time. To avoid shortchanging his employer, he built Prompt Advisers in the margins of his day: 4:30 a.m. to 9 a.m., then again from 5 p.m. to 9 p.m. As revenue grew and he hired a partner and staff, the stakes changed. Side income became payroll. He waited to see whether the AI boom would last before finally leaving his job.

By then, the business model had evolved. Early on, 70 to 80 percent of revenue came from custom AI development. But as tools like Claude Code and other advanced models matured, clients no longer wanted a black box solution; they wanted to understand how to wield the technology themselves. Prompt Advisers flipped its mix to focus on consulting and education, reserving custom builds for a handful of enterprise clients.

Today, the company runs workshops and advisory programs for small businesses and global enterprises alike, with engagements ranging from a few thousand dollars to well into five figures. The core message is consistent: clean your data and systems, understand compliance, and drop the “I’m non-technical” label. In Kashef’s community of more than a thousand entrepreneurs, many of the most active Claude Code users are people who once believed AI wasn’t for them — until they learned that the right prompts could turn a daunting technology into an everyday tool.