Crafting a perfect prompt for AI chatbots is often a challenge. This difficulty is so significant that startups are creating new roles for prompt engineers. To address this, consumer-facing AI apps are increasingly adding features like suggestion buttons and autogenerated recommendations. These features nudge customers to use the chatbot more frequently and show them what the app can do.
Hero, a productivity startup founded by former Meta employees, announced a new autocompletion software development kit today. This kit will fill in prompts for you based on context. The software development kit is currently invite-only and will let developers integrate this technology into their apps.
Here is how it works. When you are booking a flight, you might write “Book a flight” in the prompt. The autocomplete software development kit will then start populating fields like destination, origin, date, time, airline, and return date. You can stop at any point and send the query to the chatbot.
This technology is also useful when using an AI-powered image or video generator. In these cases, autocomplete can help with different parameters like object, style, location, landscape, and camera angle. Notably, Adobe has a new feature for creating soundtracks with its Firefly app. This feature lets users enter keywords in different sections of a prompt, such as mood, style, and purpose, to easily create a tune.
Hero engineer Saharsh Vedi, who worked on this feature, explained that you usually need a lot of back-and-forth with an AI app to get the results you want. With this autocomplete feature, you can achieve your goal with fewer prompts, or even just one.
The startup’s co-founder, Brad Kowalk, told TechCrunch that the company is using a series of models to predict what you might type into a prompt next. He stated that with AI autocomplete, they pull forward all the inputs needed to complete an action, finishing it ten times faster as there are fewer back-and-forths involved. He said this unlocks a whole new set of use cases ranging from travel to commerce, and from ads to customer support.
He added that because there are fewer messages involved, companies operating at scale can save on server costs.
Kowalk said that he and his co-founder, Seung W. Lee, thought of building the technology based on their experience working on augmented reality features at Meta. On AR glasses, there are constraints on screen size, so the interface for prompts needs to be simple, like adding parameters to a query.
The startup, which raised four million dollars in seed funding last year, has secured three million dollars in additional funding led by Forerunner Ventures. Kowalk said that in the coming months, depending on the app and software development kit growth, Hero plans to raise a bigger round.
Hero is testing a version of this technology in its own app. This version helps users find time for a meeting or to catch up with a friend through autocomplete prompts. The feature will be released to users in a couple of months.
The company is also in talks with Koah Labs, an ad tech startup, to build AI-powered ads where brands can show up in autocomplete suggestions.