
Adrian Holovaty, the founder of the music-teaching platform Soundslice, recently unraveled a perplexing issue that had been troubling him for weeks. Odd images from ChatGPT sessions were frequently uploaded to his site, leading him to discover that the AI was inadvertently promoting his app while simultaneously misrepresenting its capabilities. Best known for co-creating the open-source Django project, Holovaty launched Soundslice in 2012, a platform that has remained ‘proudly bootstrapped’ ever since. His current dual focus on music as an artist and as a founder has kept him busy. Soundslice stands out in the music education landscape with its innovative video player that syncs with music notations, guiding users on how to play specific notes. A standout feature is the “sheet music scanner,” which utilizes AI to transform uploaded images of printed sheet music into interactive scores. The trouble began when Holovaty noticed unusual entries in the error logs tied to the scanning feature. Rather than the expected sheet music images, the logs were filled with screenshots of ASCII tablature and text from ChatGPT. This basic text-based guitar notation system, relying on standard keyboard characters, was not what the scanning tool was designed to handle. Although the volume of these uploads didn't strain the app's resources, Holovaty found himself perplexed. He documented his confusion in a blog post, expressing his surprise at the influx of ASCII tablature images. After experimenting with ChatGPT himself, he realized the AI was misleading users into believing they could upload these sessions to hear the corresponding music through Soundslice. This revelation posed a new challenge for Holovaty: the reputational risk of attracting users with inflated expectations. Faced with this dilemma, Holovaty and his team deliberated over potential solutions. They considered placing disclaimers across the site to clarify that the app could not convert a ChatGPT session into playable music. However, they ultimately decided to develop the feature that would allow such functionality, despite it being an unexpected pivot for the platform. Holovaty expressed his mixed feelings about the situation, acknowledging the value of adding tools that benefit users while questioning whether it was right to create features in response to misinformation. This incident may mark a pioneering case of a business adapting its offerings due to the repeated inaccuracies of an AI. Commenters on Hacker News weighed in, suggesting that this scenario mirrors the persistent issue of overenthusiastic salespeople promising more than a product can deliver, thereby compelling developers to create new features. Holovaty found this comparison to be both fitting and amusing, highlighting the unpredictable impact of AI on product development.
In a significant move within the brain-computer interface (BCI) sector, Chinese startup Gestala has successfully raised ...
TechCrunch | Mar 12, 2026, 04:35
On March 11, the Ministry of Information and Broadcasting (MIB) took decisive action against the popular messaging platf...
Business Today | Mar 12, 2026, 05:55
In a recent four-minute video message, Atlassian's CEO, Mike Cannon-Brookes, reached out to employees to discuss the com...
Business Insider | Mar 12, 2026, 02:45Atlassian has announced significant layoffs affecting around 10% of its workforce, translating to approximately 1,600 em...
Business Today | Mar 12, 2026, 05:20
Nuro, a startup from Silicon Valley backed by prominent investors including Nvidia, Uber, and Softbank, is stepping into...
TechCrunch | Mar 11, 2026, 23:35