AI Prompt Completion SDK - Boost Productivity

The Challenge of Effective AI Prompting and Emerging Solutions

Developing effective prompts for artificial intelligence chatbots can prove difficult, leading to the emergence of specialized roles like prompt engineers within startup companies.

AI applications accessible to consumers are increasingly incorporating features designed to guide users, such as suggestion buttons and automated recommendations.

Hero's Autocompletion SDK

Hero, a productivity-focused startup established by former Meta personnel, has unveiled a new autocompletion SDK.

This SDK is designed to automatically complete prompts based on the surrounding context and is currently available on an invite-only basis for developers seeking to integrate the technology into their applications.

Consider the scenario of booking a flight; initiating a prompt with “Book a flight” will trigger the SDK to populate relevant fields, including departure and arrival locations, dates, times, airline preferences, and return dates.

Users retain the ability to submit the query at any stage of the completion process.

Applications Beyond Travel

The utility of this technology extends to AI-driven image and video generation tools.

Autocomplete functionality can assist in defining parameters such as objects, artistic styles, locations, landscapes, and camera angles.

Adobe’s recent addition to its Firefly application, enabling soundtrack creation, exemplifies this, allowing users to input keywords related to mood, style, and purpose to facilitate tune generation.

Reducing Prompt Iterations

Saharsh Vedi, an engineer at Hero involved in the development of this feature, notes that achieving desired results with AI applications often requires extensive back-and-forth communication.

The autocomplete feature aims to streamline this process, potentially reducing the number of prompts needed – or even achieving the desired outcome with a single, completed prompt.

Predictive Modeling and Cost Savings

According to Hero co-founder Brad Kowalk, the company employs a series of models to anticipate the user’s next input within a prompt.

“AI autocomplete proactively presents all necessary inputs to finalize an action, accelerating completion by a factor of ten due to reduced interaction,” Kowalk explained to TechCrunch.

“This expands the possibilities across various sectors, including travel, commerce, advertising, and customer support.”

Furthermore, fewer message exchanges translate to potential cost savings in server infrastructure for companies operating at scale.

Inspired by AR Interface Constraints

Kowalk stated that the concept originated from his and co-founder Seung W. Lee’s experiences developing augmented reality features at Meta.

The limited screen real estate of AR glasses necessitates a simplified prompt interface, akin to adding parameters to a query.

Funding and Future Plans

Having secured $4 million in seed funding last year, the startup has since raised an additional $3 million led by Forerunner Ventures.

Hero intends to pursue a larger funding round in the coming months, contingent upon app adoption and SDK growth.

Currently, Hero is internally testing the technology within its own application, focusing on features that assist users in scheduling meetings or connecting with friends via autocomplete prompts.

This feature is slated for public release within a couple of months.

Discussions are also underway with Koah Labs, an ad tech company, to explore the integration of AI-powered advertisements that appear as autocomplete suggestions.