Zohar Vanunu says that using Kolbo.AI, his own AI creation workspace, he made an AI video series that got 59 million views on Instagram in 30 days, and that his own account plus the product account together gained more than 200,000 followers. By the usual math, this should be a business. But by his account, these views brought in almost no revenue.[1]
What actually brought users was something else. According to his own account, he made an English-language tutorial that laid out every prompt, every image, and every version of the video from the entire project for people to copy. On the day the tutorial went live, signups from outside Israel jumped to 6.5 times; over the following three weeks, 3,285 overseas users signed up, accounting for 78% of all new signups in the same period.[1]
His product is called Kolbo.AI, a creation workspace that packs chat, images, video, music, voiceover, 3D, documents, review, and editing plugins into the same project and the same context. The website describes it as replacing the back-and-forth switching among multiple AI tools.[2]
A Single Clip Means Moving Files Between Five Websites
For example, someone wants to make a thirty-second short film. They first have to talk out the script in a chat tool, then go to an image model to generate key visuals, then take the images to a video model to turn them into footage, find a separate music model for the score, and find a voiceover tool to read the narration. The output of every step has to be downloaded and re-uploaded, and they have to keep an eye on the character's appearance to keep it consistent.
Kolbo takes on that middle stretch of shuttling. The website says the workspace has more than 100 models from external labs, including chat, image, video, music, and voice; users can let Auto mode choose, or send the same prompt to several models at once to compare. For example, the image models include Midjourney, the video models include Kling, Sora, and Veo, music includes Suno, and voice includes ElevenLabs. Projects, consistent characters, and team-shared credits all sit in one place.[2]
It does not require people to leave their editing software. The plugins run as native panels inside Premiere Pro, After Effects, and DaVinci Resolve, and generated images, videos, voiceovers, sound, or motion graphics land directly on the timeline; the plugins can be installed for free on any plan (including the free tier), but actual generation still deducts credits.Plugin documentation[3] The website FAQ also notes that DaVinci Resolve requires the paid version; this is DaVinci's rule.[2]
One rule is small but lowers the cost of repeated trial and error: failed generations themselves do not deduct credits. The website puts this sentence alongside “payment corresponds to completed work”; the troubleshooting documentation also repeats that failures do not deduct credits.[2] · [4]
Image: Official Kolbo.AI showcase[2], retrieved 2026-11-11; original interface preserved.
The Product Is the Shape His Craft Took When It Was Commoditized
Vanunu says he spent six years doing directing, editing, and 3D art, was entirely self-taught, had no programming background, later taught himself to write software from scratch, built the first version alone, and had no outside funding or team.[1]
The turning point was concrete: the first time he saw an image generated by Midjourney, he realized that the lighting, 3D, and compositing he had spent years learning had been commoditized to the point that anyone could get them in seconds; within months, his creative business dried up. He did not choose to fight generative AI; instead, he turned to building tools.[1]
That explains why the product looks the way it does. The first version was built around his own filmmaking workflow: write, generate images, turn them into video, add music and voiceover. It is a lifeboat an editor built for himself—not by moving the editing timeline to the cloud, but by gathering the generation steps before and after the timeline into one project.
Growth was also not the work of a marketing team. He had no big launch day, no budget, and no team. His approach was to first use his own product to make something people wanted to see, post only the results, not selling points or feature lists, then open the files and teach the entire production process. The earliest customers were creators who saw his work and wanted to make the same thing. He says the second channel was affiliates: customers repeated this show-and-teach with their own audiences and shared revenue when it brought in income; one affiliate brought in more than $30,000. These numbers are also the founder's own account.[1]
How Credits Are Charged, and Who It Wants to Catch
On the public price list (retrieved 2026-11-14), Lite monthly currently charges $18.90/month, with a base renewal of $27/month, and allocates 2,500 credits per month; annual currently charges $226.80, with a base annual renewal of $259.20, and credits are still distributed monthly, but the money must be paid upfront.[5]
What is really interesting is the rollover rule: within the paid access period, unused subscription credits can be kept, and the subscription bucket can accumulate up to three times the monthly allocation; annual users get the same treatment. This is a balance cap, not an expiration after three months. After canceling a subscription or when the paid period ends, subscription credits return to the free allowance; one-time credit packs never expire and remain usable after cancellation.Price list[5] By comparison, Krea's public documentation says unused compute units in individual plans do not roll over, business/enterprise monthly can roll to the next month, annual can roll over at most 12 months, and separately purchased packs must be used within 90 days.Krea documentation[6] The two companies' credits are not the same unit, and there is no cost comparison for the same task and same output, so this only shows that the rules differ; it does not show who is cheaper.
On the team side, he chose another path: no seat fees; you pay in credits, and the organization gets a shared credit pool, unlimited members, branded login, and a real-time usage dashboard.Kolbo.AI[2] For a small studio, that is easier to budget than per-head pricing, even if the total is not necessarily lower.
Between $60,000 and 730 Subscriptions
About ten months after launch, Vanunu says he makes about $60,000 in monthly revenue. This is his own framing; it does not say whether it is MRR or another revenue measure, or whether refunds and costs have been deducted.[1]
TrustMRR updated its record at 2026-11-14 12:32 AM and gives metrics from another point in time: Kolbo.AI's all-time revenue was $414,644, MRR was $46,954, and it says revenue was verified via Stripe API key; the same page separately lists 730 active subscriptions.TrustMRR[7] This is not the same point in time or the same metric as the founder's monthly revenue statement, so the two cannot be directly subtracted. Active subscriptions do not equal customers, nor do they imply plan distribution and retention. The costs of procurement, generation, and support are not disclosed, so profit remains unclear.
In the interview, Vanunu describes customers as creators, small studios, and marketing teams—people who previously switched back and forth among ten or so AI subscriptions.[1]
Kolbo's shape comes from a loss: after his craft was commoditized, he gathered his own production process into one workspace. That explains better than view counts why some people are willing to follow along—users see not a feature list but a production path they can follow all the way through.