HeyClicky got 15 million views for arguing against itself

Farza Majeed hit record on an experiment: himself working through DaVinci Resolve, guided by a tiny AI assistant that hovered beside his cursor and pointed out the exact buttons to press. The clip went viral.

Instead of turning the sudden attention into a stealth rollout, Farza published the whole codebase to GitHub the very next morning under an MIT license.

He handed over the keys, asked other developers to tinker with it, and invited the internet to help build the future of AI interfaces out in the open.

That clip became HeyClicky, and it did around 3 million views on its own. Across Instagram, X and LinkedIn, everything he has put out about the product has now passed 15 million.

Farza built buildspace into a $10 million raise led by a16z at a $100 million valuation. He shut it down in August 2024 with two years of runway still in the bank, admitting he had run out of ideas and the desire to push was no longer burning as it once did.

Clicky is the comeback, built inside Y Combinator's Spring 2026 batch. It sits by your cursor, watches your screen when you press a hotkey, talks back, and points at whatever you're looking for.

The launch post on April 7 reads like a recommendation rather than an announcement:

The appeal was pure friction reduction. Instead of tabbing out to watch a one-hour walkthrough on color grading, Farza has his project open in DaVinci Resolve, asks how to color grade the shot, and lets Clicky point out the bottom toolbar and node settings. He learns by doing in real time, and this interaction became the demo.

It did roughly 3 million views, landed at #6 Product of the Day on Product Hunt, and the code has since collected 7.3k stars on GitHub.

The clip works because nobody has to explain the product to you while you watch it. That is a low bar that a great deal of enterprise software still cannot clear.

It executes on a thesis that keeps surfacing in enterprise tooling. Sam Altman called voice and video “the best computer interface I've ever used”, writing that talking to a computer “has never felt really natural for me; now it does.”

Microsoft CEO Satya Nadella outlined this exact shift in Microsoft's Work Trend Index Annual Report, focusing on the concept of digital debt and arguing that modern software must eliminate the constant cognitive tax of context-switching between thinking, searching, and doing.

Clicky turns that thesis into a three-second visual. Resolving a software roadblock drops from a two-minute external tutorial detour to an eight-second in-flow resolution. When assistance becomes that simple, user behavior fundamentally shifts, and people ask for help significantly earlier and far more often.

Then there's the pronoun. Every post is “I built this thing,” “I've been using it,” “I'm gonna keep hacking on Clicky.” The company account took a back seat early on, though he posts from both handles today.

On day five, with the initial clip still picking up views, Farza posted a cold self-assessment: “It's not a 10/10 idea, but, I think it's an okay starting point to continue iterating into an actual company.”

Most marketing playbooks demand unyielding confidence during launch week. Farza went the other way, openly admitting to millions of people that his viral prototype was merely average.

He later made a metrics video whose stated premise is the question “is my product trash?” He opens generously, with twenty thousand downloads, half a million messages sent and ten thousand monthly users. Then he takes his own case apart:

Instead of showing a smoothed cumulative graph, he pulled up a choppy weekly active user chart settling at ten percent weekly growth.

He skips testimonials too. The users he names are kids building games, artists learning human anatomy in Blender, and founders pointing it at Google Slides “to roast their pitch.”

Farza pretty much treats his feed like a scratchpad. His teasers are low-key: usually just one sentence, a heads-up that something drops tomorrow, and a link thrown into the replies.

On April 22: "Something new. See you tomorrow, link below." On May 13: "Bringing back the computer we fell in love with. See you tomorrow, link below." He used the same see you tomorrow shape back in July 2025 for a different product entirely.

Since the updates sound like someone casually talking from their desk, people show up simply because an actual human asked them to take a look.

Farza spreads his reach across multiple channels, not just a single feed. He runs a YouTube channel under the same handle, has used livestreams as a launch format, cross-posts to LinkedIn, and the company's own Y Combinator page claims a fourth channel outright: “We've gone super viral on Instagram and are getting insane pull.”

Between the short demo clips, the long-form video, the livestreams and the written posts, he is running something closer to a small media operation than a launch campaign.

The core lesson from Clicky's rollout is that the most powerful marketing is the content a company never produces itself.

This is the exact organic engine that fueled ChatGPT's initial explosion. When OpenAI launched, it did not rely on ad campaigns or polished product marketing. The distribution came from millions of users showing off their own experiments, prompt structures and weird edge cases.

We saw the same thing in our recent breakdown of World Labs' Atlas launch. Just as Fei-Fei Li's announcement gained its real momentum when users began turning personal photographs into walkable 3D worlds, Clicky took off because people immediately started showing off what they built with it.

The same pattern showed up days ago with Meta’s Muse, its personal AI agent that “gets things done across every part of life.” The official launch post (backed by a demo video and the product site at muse.ai) crossed 5.8 million views on X within roughly 48 hours, while Meta’s own announcement framed the product around concrete, in-flow actions rather than abstract capabilities. When the value is visible in seconds, even a large-company launch rides the same organic wave Clicky did.

By publishing the codebase under an MIT License on GitHub on day two, Farza let the internet run the campaign:

When an interface makes utility immediately obvious, users become the marketing department. The product only needs to give them something clear enough to show off their own work.

In June 2025 Farza launched Tidbit, a better interface for ChatGPT. Four days in, he posted a progress report that reads like a rehearsal for the Clicky video:

Eight weeks later he put a public deadline on it in a post titled “Tidbit's fate: 3 weeks to prove its worth.”

You have almost certainly never heard of Tidbit.

He had exactly the same audience for Tidbit that he had for Clicky, and it made no difference at all. The thing that separated them was whether a stranger could tell what the product did by watching someone use it, and a better interface for ChatGPT does not really show up on camera.

Most coverage settles on the easy explanation that Farza already had an audience (DailyDropout wrote that it “didn't need a marketing budget; it needed one good tweet”). But Tidbit had that exact same audience and died anyway.

There's a bill attached to working this fast in public, too. On July 6 the team shipped proactive agents, software that acts on your screen activity without being asked. Twenty-six days later the changelog read: “Proactive agents are off, and all the local activity tracking that powered them is gone.”

Withdrawn over privacy, disclosed by the company itself. Farza settles that account in one changelog line. Anyone running communications at a public company knows how differently that conversation goes in their building, and that gap is the honest boundary around everything above.

Try this at your next product review. Grab someone who’s never seen the product, play them thirty seconds of silent screen capture, and ask what it does. If they can’t answer, a founder video won’t help. The issue is earlier than marketing. Farza’s launch clip was that test, and it worked the first time.