The war against clunky AI prompts just got a new weapon. Hero, a productivity startup founded by former Meta engineers, launched an invite-only SDK today that autocompletes AI prompts based on context - promising to slash the endless back-and-forth that plagues current chatbot interactions. The move comes as companies scramble to make AI interfaces more intuitive, with Adobe recently rolling out similar structured prompting features.
Hero is betting that the future of AI interaction isn't about better models - it's about better prompts. The productivity startup, founded by former Meta AR engineers Brad Kowalk and Seung W. Lee, just unveiled an autocomplete SDK that fills in AI prompts as users type, potentially ending the frustrating cycle of prompt refinement that's become the bane of AI adoption.
The technology works like predictive text on steroids. Type "Book a flight" and Hero's SDK starts populating fields like destination, departure date, airline preference, and return timing. Users can stop at any point and fire off the query, or let the system build out a comprehensive prompt automatically.
"With AI autocomplete, we pull forward all the inputs needed to complete an action, finishing it 10 times faster as there are fewer back-and-forths involved," Kowalk told TechCrunch. "This unlocks a whole new set of use cases ranging from travel to commerce, and ads to customer support."
The timing couldn't be better. As AI chatbots proliferate across every app and service, users are hitting a wall with prompt engineering. Companies are hiring dedicated prompt engineers, while consumer apps pile on suggestion buttons and pre-written examples to guide users. Adobe recently tackled this with its Firefly soundtrack generator, breaking prompts into structured sections for mood, style, and purpose.
Hero's approach goes deeper, using what the company calls "a series of models" to predict the next logical parameter in any given context. For image generation, that might mean suggesting object, style, location, landscape, and camera angle based on an initial seed phrase. For customer service, it could populate common complaint categories and resolution paths.
The inspiration came from an unexpected place - augmented reality. Both founders worked on AR features at Meta, where screen real estate constraints forced them to rethink interface design. "On AR glasses, there are constraints on screen size, so the interface for prompts needs to be simple - like adding parameters to a query," Kowalk explained.
