YC-backed productivity app Superpowered pivots to become a voice API platform for bots

Calendar apps are important for productivity but it’s hard to differentiate enough to sustain growth from core usage. Backed by Y Combinator Superpoweredwhich is an AI-powered notetaker for your meetings that doesn’t involve recording bots, it hit a road block and is now pivoting to become Vapeis an API provider so that anyone can easily create a natural-sounding voice-based AI-powered assistant.

Superpowered was founded in 2020 by Jordan Dearsley and Nikhil Gupta. But after three years of working on it, Dearsley said the team wanted to work on a more challenging product. The company has not closed the initial product because the startup says that Superpowered is profitable – it is in the process of bringing someone to run it. Y Combinator said in June that more than 10,000 people were using the product every week, but the company did not provide any updated numbers.

Image Credits: Vape

To date, Superpowered/Vapi has raised $2.1 million in seed money from investors including Kleiner Perkins and Abstract Ventures.

Pivot to Vapi

The company offers Vapi as an API to allow developers to create a bot using prompts – after putting it behind a phone number. Additionally, it offers an SDK integration so developers can embed the bot in websites and mobile apps.

Dearsley told TechCrunch via email that the idea to create Vapi came from a personal problem. He moved to San Fransisco and began to miss his friends and family, who were in a different time zone. He created an AI bot attached to a phone number on the other end to talk to a human to sort out his thoughts.

“I like it, but I’m always disappointed by its unnaturalness. It’s not like talking to someone. The voice is buzzing, there are long delays before it responds, and it interrupts me while I’m talking.” he said.

“So I kept working on it and walking with it. Eventually, we were drawn to this problem of conversation. It’s really hard to make people feel something. Voice assistants today are clunky and turn-based, we want to build something that feels human.

Technically, Vapi currently integrates a set of third-party APIs to create a robust voice conversation platform. For example, it uses solutions from Twilio for telephony, Deepgram for transcription, Daily for audio streaming, OpenAI for answers, and PlayHT for text-to-speech.

ScaleConvo, a startup in the YC winter batch for 2024, is already using Vapi to launch conversational bots for sales teams and property management companies. However, Vapi did not disclose its other clients. The company has opened up its API to Vapi Phone and Vapi Web products now.

Challenges for Vapi

One of the biggest challenges for startups is reducing latency, according to Magnus Revan, an ex-Gartner analyst and chief product officer at multimodal conversation startup Openstream.ai.

“OpenAI models need between 2-10 seconds to generate a response – while on the phone the gold standard is 700ms between the user finishing speaking and then the ‘bot’ starting speaking. And going of sub 1-second latency with capable models (high parameter count of open-source models like LLaMA2 70B) is very difficult,” said Revan.

Currently, Vapi has a latency of 1.2-2 seconds depending on various factors. Dearsley expects to drop latency below one second next month thanks to Vapi’s own work and OpenAI improvements.

Mohamed Musbah, an angel investor in Vapi also said that the startup solution will improve the overall development of API.

“As OpenAI and others improve their models, Vapi’s platform will become more powerful, equipped with better knowledge bases, code execution capabilities, and more context windows. The focus of Vapi in solving the biggest friction areas in voice communication is its content as user demand grows for voice assistants,” he said.

However, this puts the onus of improvement on other solutions rather than Vapi itself. Dearsley says that reliance on other APIs will reduce Vapi’s vulnerability if larger companies start moving into that area. However, the team says it has an edge in terms of building the infrastructure to handle thousands of calls simultaneously. Dearsley emphasized that as Vapi’s web and phone APIs roll out to the public, the team will also look to build its own models for audio-to-audio solutions.

Leave a comment