From the Diary of Bassam Nouh

For the owner who knows AI could save time and doesn’t know where to start: hire it the way you’d hire a person, one job at a time.

Every step in the workflow showed green. The automation had one job, to email its owner a brief every Monday morning, and by every signal on the screen it was doing that job. The first two emails arrived completely empty. “A test setting fed dummy data downstream, so the system worked perfectly on nothing,” says the man who built it. “Most SMEs would have seen the green lights, called it done, and never read the output.”

It is a small case of a common pattern. By some estimates, more than 80 percent of AI projects fail, twice the rate of IT projects that do not involve AI, researchers at the RAND Corporation reported in 2024. One root cause they identified is an organization that “focuses more on using the latest and greatest technology than on solving real problems.”

Bassam Nouh has a plainer way of saying it. “They start with a tool rather than a process,” he says of small and medium-sized businesses trying AI. “They buy a chatbot, then look for something for it to do.” He is the founder and CEO of Digital Medium, a done-for-you AI automation agency that builds AI employees and agents for SMEs, and his career before it was in HR and recruiting. The Monday brief was his own build. It took 95 minutes to make and costs about 2 cents per run.

Leader in Spotlight: Read Brian Gould Feature in Executives Diary

Where the Automation Goes Wrong

Bassam is specific about who he is writing for. He names SME owners and “the people who run their sales, support, and recruiting teams,” and he describes them in one line: “They know AI could save them time but don’t know where to start.”

His account of what happens next is a short list. The business starts with the tool. Then, he says, “Nobody writes down how the work is really done today. The automation copies the official version of the job, not the real one.” They try to automate everything at once “when they should start with one job that eats hours every week.” And they forget the people: “Staff who weren’t consulted quietly work around the new system.”

Process before tool is also where the survey evidence points. In McKinsey’s 2025 global survey on AI, the redesign of workflows had the biggest effect, of 25 attributes tested, on whether an organization saw an earnings impact from generative AI. Twenty-one percent of respondents whose organizations use it said those organizations “have fundamentally redesigned at least some workflows.”

The last item on Bassam’s list is the Monday brief again: “They trust the dashboard instead of the output.” What he took from two empty emails is now, he says, “the lesson I give every client: judge an automation by what it produces, not by whether it runs.”

He says he learned the gap between the written job and the real one somewhere else.

A Career Spent Hiring

Before Digital Medium, Bassam’s work was people. The roles he asks the piece to name run back to 1999: Personnel and Administration Manager at M. H. Alshaya Co., Human Resources and Administration Manager at Carrefour, Group HR Director at EGTC Group, and founder of KZA Consult in 2016, where his profile lists executive search, OKR systems and performance management as his areas of expertise. Along the way he took an MBA in Human Resources at the University of Leicester, and at M. H. Alshaya Co., as he puts it on his profile, he “Programmed myself an administration management software” for visas and residences, labor cards, health cards and trade licenses. He also lists the H.H. Sheikh Khalifa Excellency Award, which he dates to 2004.

Asked for the person who shaped how he works, he named a book instead: Geoff Smart’s Who: The A Method for Hiring. “It taught me to hire against a scorecard of outcomes rather than a job title,” he says. “I now build every AI employee the same way.”

What that career left him with is a short list.

What Hiring Taught Him About Automation

He applies the list to software as it stands. “An automation is a hire,” he says. “It needs a clear job description, a manager, a probation period, and a performance review. Most failed projects skip all four.”

“Processes live in people, not documents. After years of interviewing and managing teams, I know the real job is always different from the written one. Engineers automate the written one.”

Bassam Nouh, Founder and CEO, Digital Medium

The other two lessons are about the team around the system. “Adoption is a people problem,” he says. “If the team feels replaced, they’ll make the system fail. If they feel it takes away the work they hate, they’ll champion it.” And years of OKR and performance work, he says, taught him to agree on what “better” means in numbers before anything is built.

A project with Digital Medium follows the same order.

How a Digital Medium Project Runs

“I start the way I used to start a recruitment brief,” Bassam says. “I sit with the person who does the work and write a job description: what they do step by step, the judgment calls, and the workarounds nobody documented.”

Then he picks one job to hand over. “It should be repetitive, measurable, and expensive in hours, like answering leads, screening applicants, or sorting an inbox.” He calls this hiring one AI employee.

He builds on what is already there, “the calendar, inbox, and CRM the business already has,” and nothing goes out unchecked at first. “A human approves anything that goes out until the system has earned trust, like a probation period.” Only once that role runs reliably does he add another.

The simplest version, he says, is the AI Chief of Staff: “one project, one charter page and four job prompts, running on a $20 monthly subscription with no code. It reads the week, preps every call, finds threads that went silent, and writes the follow-up.”

That build, like the others, is public.

What He Publishes, Including What Broke

Asked for evidence, Bassam points to the Monday brief. On its first run, he says, it found 19 email threads that had gone silent without a reply, the oldest 63 days old. “Those were real conversations with real people that had simply fallen through the cracks,” he says. “All of it is published, including what broke.”

“I build in public. Every system I make for myself becomes a video, a newsletter issue, and a free template, including the parts that broke.”

Bassam, Founder and CEO, Digital Medium

His own list of what is live today has four entries: Digital Medium’s AI automation services, KeyText, an AI text expander for Chrome, KeyWeek, a weekly agenda for tasks and calendar, and Build Leaner, his YouTube channel and weekly newsletter, Steal This System, with free importable automation templates. LavaBella, a beauty iOS app, is still being built. They are built in public, he says, so they “show exactly how I work.”

One of his builds is a board of AI advisors, a CFO, a risk officer and a marketer, told to argue before he spends anything. “You’d fire a human advisor who agreed with everything,” he wrote. “So why does your AI still have the job?”

His hard moment, when asked for one, was a rejection. Google turned KeyText, his Chrome extension, away from the Chrome Web Store. “I didn’t argue with the review,” he says. “I went through the code line by line, found a permission we had requested but never used, removed it, and resubmitted. Now every build I ship asks for only what it actually uses.”

Where Bassam Is Now

Bassam is in Ras Al Khaimah, in the UAE, and Digital Medium is based in Dubai. His main goal for the next twelve months is “more SME clients for Digital Medium,” with Build Leaner, KeyText and KeyWeek put in front of the same readers.

The readers he wants to hear from are the ones he described at the start: SME owners and the people who run their sales, support and recruiting teams, with one job in mind that eats hours every week. He asks that no client names appear, and none do.

Outside the agency, he helps run his family’s two med spas, one in Florida and one in Dubai. “That’s where most of my systems get tested first,” he says, “on real bookings and real clients.”

What’s next, in his words: “Building a team of AI employees that any SME can hire one at a time, and showing every build in public.”

The Bassam Nouh Playbook: Hiring an AI Employee

Write the job description first. Sit with the person who does the work and record the steps, the judgment calls and the workarounds nobody documented.

Hire one AI employee at a time. Pick one job that is repetitive, measurable and expensive in hours, and add the next only once the first runs reliably.

Build on the tools the business already has. The calendar, the inbox and the CRM come first.

Put every new system on probation. A human approves anything that goes out until the system has earned trust.

Judge an automation by what it produces, not by whether it runs. Read the output, and agree on what “better” means in numbers before anything is built.