Tech demos are the theater of Silicon Valley. Within the Valley, there’s only a limited amount of attention — and time — to parse through the hundreds of product launches every week. On X, the stage, founders fight for virality. Few manage to get attention, fewer outlive the initial wave of hype, and even fewer sustain attention week after week.

Not so for Farza.

Over the last few months, the founder of an AI company that’s not yet a household name has become something of a recurring character on X. When he posts, he appears center stage with a laptop in seemingly impromptu demos, speaking with conversational casualness, accompanied by game-play music.

In one such demo posted on X in April 2026, Farza introduced an AI companion, HeyClicky. The companion, a small blue triangle resembling a play button, lives right next to his on-screen cursor and teaches him how to use DaVinci Resolve, video-editing software, by taking full hands-free control of his Mac, and talking him through it; it walks him through color-grading, advising on which data file to employ for coloring based on the type of raw file the video uses. “I don’t wanna watch a one hour YouTube video about it [video editing], I wanna learn by doing,” Farza explains. The post hit almost 3,000,000 views, with 5,000 people signing up for access to HeyClicky. This was unexpected for Farza, who had just been playing around with what he thought was a “dumb idea.”

In other demos, Farza summons HeyClicky (by saying “Hey, Clicky”) with the instruction to help learn the Pythagorean theorem. HeyClicky — with access to his entire screen — draws precise polygons, points with pixel-perfect accuracy at specific UI elements, and walks him through complex steps in real software, instructing as it goes. In another, he asks HeyClicky to turn on some music, the AI opens Farza’s Spotify app, plays the requested song, only for Farza to ask to lower the volume — which it obliges. All of these posts consistently reach millions of organic viewers in a matter of hours.

Farza is the internet nom-de-plume of Farzain Majeed. He’s a software engineer, entrepreneur and educator. More importantly, Farza is doing something that every technology founder aspires to: convincing ordinary people that AI is magical, and showing them how. For the average American, the capabilities of AI have felt limited since the late 2022 launch of ChatGPT. Casual users might ask ChatGPT to write an email, come up with a recipe, summarize a document, or plan a vacation. Most Americans remain cautious. We have been told, repeatedly, that AI is going to change everything, but “everything” is difficult to picture; the distance between what AI companies promise their technology can do and what people actually experience still feels enormous. Farza is showcasing an unusual talent for closing that distance.

Much of what HeyClicky demonstrates — screen comprehension, voice agents, narrative guidance, drawing and pointing — has been discussed, promised, or demoed elsewhere. The underlying capabilities were developed by some of the world’s most valuable technology companies: HeyClicky assembles reasoning from Claude, text-to-speech from ElevenLabs, and voice transcription from AssemblyAI into a single experience.

What Farza seems to understand is this: the most powerful technology in the world is not, on its own, enough to compel people.

Farza became an entrepreneur because there was no other option. His parents told him that he wasn’t allowed to get a job. Born in Lahore, Pakistan, Farza moved to Pembroke Pines, a suburb in south Florida, aged five. “My dad got a job making pizzas, and then he got a job selling cars at Nissan,” he says of his family’s arrival in 2001. “My mom took care of us three boys all the time. I’m the oldest.”

Farza received his first computer when he was nine, an Acer AST180, which was his introduction to the internet and to multiplayer online games like Club Penguin. He received his first job offers when he was 13 — a grocery bagger, and then a sign spinner outside a gym. “It was $15 an hour, which is huge in 2008 or 2009.” But his father forbade it: “you must do something yourself.” So Farza started an eBay business, first selling a copy of Dragon Ball Z: Budokai Tenkaichi 2. The business grew into selling his family’s cast-off clothes, then Sony Blu-ray discs and printable dual-layer DVDs, generating more than $100,000 in annual revenue and making Farza a top seller on the platform.

Farza recalls that his father “would always quiz me at dinner every day about what I did, why I did it, how much did I make last week? How much did I make this week? Why is it less?” Farza then walks me through a long personal history of being “unemployable.”

Having become a successful entrepreneur at 13, Farza’s only goal upon graduating high school was to just go to New York City. The addition of an iMac to the family home had piqued his interest in making movies, and with the money he had made from his eBay business, he set himself up a YouTube Channel. “I was already making pretty good money. I wasn’t supposed to go to college, but Pakistani parents, you know…” They expected him to get a degree, and so he enrolled at the University of Central Florida to study filmmaking — with the remaining profits from the eBay business footing the bill.

“I got to the school, and I found that they wouldn’t let you actually make movies or rent out the cameras until your second year,” he says. He was disillusioned: “I got really mad and thought this is nonsense and bullshit. I should be able to make movies from day one if I want to.” So he visited every class he could, looking for people who were “actually doing stuff.” He found that the business classes were “fake,” medicine took 10 years, law was boring. “But in computer science classes people had their laptops out day one — and they were making stuff.” Already deep into gaming, he decided to pursue computer science.

During his time at UCF, Farza studied under Professor Niels Lobo, helping him train models of self-driving cars. “I was number one in this class with the best test scores, and the professor was like, ‘I want you to train AI models at my lab.’”

This was in 2016, years before ChatGPT or Claude, when AI was far away from its present-day ubiquity, and Tesla had just released its Model 3, with Elon Musk predicting that true autonomy was only a few years away. “I ended up being around some really smart people by accident,” he explains that the school’s computer vision program grew into being its most renowned offering.

He spent two years at Lobo’s lab before deciding against a PhD, owing to the fact that it would keep him away from ‘the real world’ for too long. Instead, he released an open-source tool for League of Legends that let users analyze their own gameplay (and anyone else’s). Farza’s project went viral and was soon picked up by professional League teams. “Everyone started offering me a job, everybody you can imagine, from the Googles to the Facebooks — everybody. And I was like, no, I don’t want to do any of this. I want to do my own thing.”

One company that did catch his eye: Mayhem, which was building an AI gaming coach. “This is 2018, I spent months there training their models. It was a lot of fun. The product ended up getting banned by Blizzard.” Blizzard – the California developer behind Warcraft and Overwatch — felt that the software Farza built at Mayhem amounted to cheating, and banned it. The software watched live Overwatch gameplay and detected when opponents had spent their most powerful abilities, sensing specific character movements to reveal a player’s vulnerability. Farza defends the product on the basis of its technological accomplishment: “Even now, I look back and I think, that’s nice. I had a vision model that was running live — this is all before all the LLM stuff. So it’s all manual models trained on that one task.”

Back then, software engineers built from the foundation up: collecting millions of data points, manually labelling them by hand, designing models for narrow tasks, tuning them to run reliably. There was no shortcut, no taking a broadly capable model and teaching it a new skill with a few prompts.

Farza lasted six months at Mayhem before the urge to work for himself returned. “I get hit up by this other company in New York, and they were looking for a cofounder/CTO.” It was another computer-vision video game venture, and Farza, now in his 20s, was worried. He described himself as “broke, with no Ws under my belt, all my friends are slowly starting to succeed.” So he took the job, hoping for a win. He helped build the product and team — and, again, lasted only six months. He just didn’t feel it anymore.

“I went back to my parents’ house. 24 now. Everyone’s confused. What happened to this guy?” Farza reminisces about his abrupt return to Pembroke Pines; “I was the guy in my neighborhood that was supposed to make it, do something, and leave.”

He was sure now, as he had always been, that he wanted to found his own startup, that he wanted to make things “until my brain broke,” but nothing felt right. His cousin Sameer dropped by once a week to learn how to make Minecraft servers. “It was the most fun I had in a given week,” so Farza decided to become a full time coding tutor. He started with the neighborhood kids, distributed pamphlets, and talked to parents, but no one bit. Then Sameer, 12 at the time, said to Farza, “I wish I didn’t have to go to school.”

“I had a terrible social experience going to school,” Farza says. “Brown kid in a South Florida school after 9/11. For 10 years, that’s all I ever heard about. Also, I was fat. I was like 260 pounds.” So he took Sameer’s plea seriously. “I looked it up. Turns out there’s something called homeschooling.” According to Farza, homeschooling at the time — before Covid-era school closures forced a generation of children online – was a “weird thing” only parents “who didn’t want their kids learning about evolution” did. Few parents and educators took it seriously, many weren’t even aware that it was an option; the legal process of becoming homeschooled was, across much of the country, quite convoluted.

Farza suspected homeschooling had the potential to escape from the fringe. “When I was 13, I wanted to not go to school. I wanted to work on my company” — his first one, reselling on eBay — “because it was fun. There must be more kids like me. Turns out, there were.” He started ZipSchool, a service to simplify the legal process of homeschooling. The software collected parents’ information and documents; Farza handled the paperwork himself — contacting school districts, filing, completing the steps to withdraw a child from school. Every state had different homeschooling laws, and he figured that if he made the process easier, more parents would take it up. ZipSchool reached 50 customers in its first three months, mostly made up of parents who “felt their kids were very, very smart, and they felt that school was moving too slow. Parents who felt like the school was just not enough for their kid.” Farza told me about a few of the students, including a nine-year-old named Sanjay, a “genius” — who would have had to spend his days at school waiting for the class to catch up, but could move at his own quick pace at home.

The company quickly became more than just administrative and legal work. “The parents kept asking for more and more,” so Farza learned how to make worksheets for Singapore math, a teaching method developed by the government of Singapore in the 1980s known for its emphasis on conceptual mastery, in high demand by the parents. “I would always ask them: what are your other problems? All of a sudden, I’m running one of the biggest worksheet download systems in the world. And people love my worksheets because I have a great small team of contractors making them.” Then the parents wanted Zip to teach their kids, so Farza recruited 70 teachers for live online classes. “They would stream themselves to kids — music, space, everything that wasn’t math or reading was my thing. We had thousands of kids in a class, the average age being seven. It became huge: Twitch, but for teachers.”

At the time, ZipSchool was only charging parents for the initial services, making $500 a month in recurring revenue. The success of the project so far convinced Y Combinator to include him in the Covid-era online batch, complete with funding — then $125,000. His daily updates to his 500 followers on Twitter also caught the attention of an investor at a16z; the firm then put in $2 million.

When Farza finally introduced a paid plan for online classes at ZipSchool, the business exploded from close to $500 to $30,000 in MRR in a single day, “I had these weird feelings around charging people back then.”

Then, convinced the product wasn’t the right long-term bet for him, he refunded every customer and pivoted. “At demo day, I was with my YC partner. He’s like, “why the fuck did you refund everybody?” I’m like, ‘I don’t know.’” ZipSchool experimented with small-group online classes, a virtual classroom platform, even a math-learning game, before concluding the real problem wasn’t the product, but the nature of the niche: in children’s education, the user and the customer are different people. That realization led Farza to abandon K-12 altogether. “Let me go to where the customer controls their destiny. Because the child cannot control their destiny. ZipSchool, but for adults. What if an adult could join a system where they could essentially pursue their dreams? What would that look like?”

***

Buildspace was his answer. It started in 2021 as a continuation of ZipSchool. It was a “stupid, simple idea.” You come in on a Saturday and build your own first piece of software in a weekend, starting from a live stream of Farza teaching, and then working through the coursework he’d designed. Buildspace billed itself as the “Hogwarts for builders and dreamers,” where anyone and everyone was welcome to turn ideas into reality — without needing a traditional job path or formal education. It reached more than 125,000 people. “To this day, we are still the biggest institution where people could work on their own stuff. No accelerator or anything comes close.”

The operation was largely remote at first — most people joined Farza on livestreams from their homes across the world. But it quickly expanded to a physical campus at Fort Mason in San Francisco. “Six months after we started, we had some really good numbers. So we ended up raising $10 million, $5 million from a16z, and $5 million from other people.”

I was on the mailing list for Buildspace, and in late August 2024 I received an email from Farza announcing an abrupt end to the program. This was very confusing; Buildspace had been one of the most happening accelerators in Silicon Valley, season after season. Just weeks earlier, an email invite had gone out for demo day of Buildspace’s “largest season of all time.” The email was brief, not even a hundred words, and linked to a letter Farza had addressed to “you,” explaining that the shutdown was due to neither lack of funding nor administrative troubles. “buildspace feels ‘done’ to me,” Farza wrote in his signature lowercase style, “the fire to push isn’t there. and i’m not enjoying the work as i once did. on top of that, i don’t have a clear new direction, or vision that would take this company to the next level.”

When I ask what exactly happened at Buildspace, Farza retorts: “At what part? Because a lot happened. It was magical. We were doing really well on paper.” He describes plans in motion for campuses around the world — in Argentina, Greece, Dubai — to bring people together and to teach them to turn their passion into income. But Farza insists that the passion for the project just wasn’t there anymore: “Five years is a long time working on a company. And every day is nonstop. You grow up, you don’t know what you want anymore. Things change, motivations change, and you just change.”

In the fallout, a large technology company offered to buy some of Buildspace’s intellectual property. This deal would have let Farza announce an acquisition — a move his investors encouraged, tempting him with the familiar trappings of startup success. There would have been a celebratory blog post, congratulatory LinkedIn updates, the perception that he had “made it.” Farza refused. “I hate lies,” he says. “This company failed because I was out of gas. We’re stopping because I’m out of gas, and I failed. This is shit. And we’re moving on.”

***

“I kind of compare myself sometimes to a basketball player. I’m obviously a good draft pick. And I hurt my ankle in the playoffs. I can’t play anymore. I’m on the ground, and I can’t even get up. And now I’m a failure. In front of everybody. That sucks.” When Buildspace shut its doors, Farza was sure he would never come back to tech. At most, he was planning to “fuck off to some European country” and start an online business. He spent the next few months traveling from one random airport to the next, never planning, just hopping on whatever flight was available. He introduced himself simply as “a writer.” Strangers confided in him anyway. The conversations usually ended the same way: Farza helping them build toward whatever they wanted next. In Morocco, his surf instructor, Hassan, dreamed of opening his own surf camp but was earning just $50 a week. Farza taught him the basics of AI-assisted copywriting and web design, and together they built the landing page for Hassan’s first business.

On a December 2024 stop in Rawalpindi, Pakistan, his hometown that he visits every year or so, a conversation with his uncle finally let something settle. “He started talking about great athletes and great directors and great musicians, and how they all had horrible periods in their life where they were just complete failures. And then he asked me: are you gonna get back up again and play the next game or are you just not gonna play the game at all?”

“For me, it was always about making things,” Farza says. “Tech, movies, music — I love making things. I love making art. I love making interesting things for people.” Software was simply a conduit for this spirit. Buildspace produced code, yes, but it also commissioned original music, produced films, and treated product launches as creative performances. “My vehicle is just a company,” he says. “Software is just one of the greatest business categories in history.” Not because of the money, he argues, but because of its extraordinary leverage: one person can make something that millions experience. “Let me pick the best business that humanity has ever created, which is software,” he says. “And let me do something really special here.”

Farza spent most of 2025 just making things, from GPT life assistants to Gemini powered sports coaches to journal apps, most of their demo videos reaching millions on X. Then, on October 9, Farza posted a demo of an “ai cursor that lives on my screen as a buddy + recommends me new stuff as i use my computer.”

“Most of the things no one remembers I worked on,” Farza says. “But it was a lot of stuff, and it was actually a very painful, very hard year. But then 2026, I was like, okay, cool. I kind of don’t wanna work on random stuff anymore. Let me pick one direction to focus on.” AI was the “obvious” next move for Farza. He’d been in the field nearly a decade and believed its greatest potential lay in amplifying the kind of guidance he gave naturally. His first instinct was to go straight to his audience: he launched a live YouTube series called Make Something, teaching viewers to build their first AI applications. Hundreds of viewers became thousands, and soon he and a handful of friends were spending hours in Discord answering screenshots and debugging problems for newcomers. Farza has an unusual problem-solving method: “What’s the stupidest thing that 12-year-old me would do to solve this?... I’ll listen to 12-year-old me.”

Eventually, the “obvious” question presented itself: why should humans answer the same questions over and over when AI could do it for them? Farza created a Discord bot that reads screenshots and responds automatically. This evolved into an AI that could watch a user’s screen in real time, spotting mistakes as they happened and offering guidance — an early version of what would become HeyClicky.

The project was now also backed by YC, though Farza had little interest in another accelerator, preferring instead to spend his time teaching AI on YouTube. But Aaron Epstein, the same YC partner who had backed him seven years earlier with ZipSchool, was persistent: “just show up.” Farza relented. “I have this weird pattern of other people believing in me more than I believe in myself,” he says. “I’m kind of self-defeating sometimes.” When he did show up, his batchmates gathered around him,expecting him to be their YC partner, “I had to tell them I was doing this with them. I’m working on a YouTube channel.”

Three weeks into the program, while exploring an entirely different idea, the YouTube channel, Farza found himself trying to learn the video-editing software DaVinci Resolve. Frustrated by the official tutorials, he repurposed his earlier prototype he’d built to help beginners learn AI. Within an hour, he had a system that could watch his screen in real time, answer questions, and guide him through the software as he worked.

He used it privately for weeks before showing it to Epstein, who urged him to publish a demo. Farza dismissed it as “a dumb idea.” The video went viral. When I ask about the production of the demo, he told me that “I recorded it right before a YC dinner in like, 10 minutes. I just had to get this out and move on with my life. Because to me, it’s like an idea is like a book on the bookshelf, and if you don’t take it off the bookshelf, uhh, it’s just gonna get dusty.”

Only then did the prototype begin to resemble a product. “I put four homies together for 30 days to help me push product because I’m not good alone,” Farza says. They developed HeyClicky, an AI desktop buddy that sits next to your cursor on Mac. It sees your entire screen in real time, lets you talk to it by voice, and speaks back while pointing, highlighting, or walking you through anything on-screen.

“Many things go viral,” Farza admits, but virality alone has never been a sufficient reason to continue working on a project. HeyClicky, he claims, felt different because of what happened after the original virality faded. On the day before demo day, he added analytics — as an afterthought — and what he found was a small but unusually engaged group: around 60 or 70 people who had returned to the product for six consecutive days. “I was like, whoa,” he says. “Sure, 5,000 downloads, only 70 retained — but who are these 70 people?”

He looked closer. They were using HeyClicky in ways he had never anticipated: children learning to build games, Shopify users improving their stores, designers working through Figma, employees asking for feedback on messages to their bosses. For Farza, this was the signal. These users weren’t using HeyClicky because they were curious about AI; they were using it because it made them more capable. “I think about the 45-year-old nurse who downloads my product at 3 AM when she comes home because she saw it on Instagram. I think about her face. How would she see this? How would she read this? These are all things that people were not able to do before,” he says. “No product has ever allowed them to do this.”

I pointed out that, despite a lifetime of entrepreneurship, HeyClicky is Farza’s first real institutionalized product. He charges $20/month for daily users, and $100/month for “power users,” who need 1,000 agent messages a month. Farza agrees: this time, he told me, he wanted to flex his business muscle. “I feel like I’m one of the best business people in the world, but I’ve never been able to flex that.” He attributes that partially to the type of companies he’s worked on before: an education company, a creator-focused platform — businesses he was proud of, but ones he felt didn’t fully exercise his commercial instincts. Yet his old hesitation for charging users remained. When his “homie” Kamil pushed him to introduce paid plans, Farza resisted. “You’re being fucking stupid right now,” Kamil told him. “You need to charge money.” Eventually, Farza agreed.

***

Farza bristles at my suggestion that HeyClicky is an education company. “Absolutely zero. Negative 10 percent of me wants to work in education,” he says, pointing to YouTube. “People say YouTube is the best education product. It’s not. It’s the world’s best software for distributing and monetizing video. It just so happens that education is one of the things people use it for.” Rather than building exclusively for education, he wants to build a better interface between humans and AI. If HeyClicky’s interface works, he believes learning will emerge naturally — as will work, creativity and countless other use cases. “I want to build a new interface for the models that makes it easier for the average person to interact with them and get more value out of them.” The computer, he reminds me, was not created to educate people but to create spreadsheets. Increased human interaction with computers just happened to unlock everything else: email, the world wide web, social media, video streaming, online shopping, and now, AI itself. What HeyClicky intends to do is build this interface for AI, the next step in the evolution of how humans interact with machines, and prosper in the process of doing so. Rather than calling itself an AI tutor or education platform, HeyClicky is sticking to its function — a way for ordinary people, anywhere in the world, to interact with powerful AI models and extract real value. “No one’s built it yet,” he says. “If AI is the most powerful chip in the world, no one’s really built its laptop. We’re still using AI on terminals. So it feels like the actual ‘70s. We’re literally accepting that the terminal is the best interface for this super magical technology called AI.”

He then explains how he can’t play the guitar, can’t sing a song, but that he can create experiences that make him — and others — feel the same emotions that art has always evoked in him. “At Buildspace, the amount of people I saw in tears watching some of our stuff. How many things have you seen or products have made you cry? Mostly movies. But products? I had one woman messaging me, ‘I literally cried watching HeyClicky’s onboarding. The way you popped off my screen, the way you said goodbye, the music, I literally cried. And it gave me hope.’” And I just couldn’t believe it. So I’m going for emotion all the time. That is the through line of what I do. I’m always asking the team, how do we feel when we use this? How do we feel in the morning when it talks back to us? How do we feel when we look at the landing page? Does this feel scammy? Everything is about how we feel.”

One of Farza’s shorter-term projects from a few years ago captures this philosophy better than any other. He never named it; there’s just the GitHub url, github.com/farzaa/ai-movie. It is essentially a computer-vision experiment in which AI observes and describes the world around him. Farza posted it to his X, with the caption “It will…chronicle my life as if I’m a heroic character in a retro-style, epic movie. All while playing Hans Zimmer in the background baby.” To me, this feels like the purest expression of his creative worldview: life is a movie, and the computer and its tools — for health optimization, music production, video editing or social media — become collaborators in the act of living. The computer is simultaneously a witness, a creative partner and, increasingly, a companion.