Cohere Wants to Build the Definitive NLP Platform

Beyond generative models like GPT-3.

Alberto Romero
8 min readFeb 21, 2022
Cohere logo (credit: Cohere)

The NLP explosion took off when Google published the seminal paper “Attention Is All You Need,” proposing the now-famous transformer architecture as a means to simplify language models (LMs) by removing the necessity for recurrence and convolution. Aidan Gomez, who was working for Google Brain at the time, was one of the co-authors of the paper. He and his colleagues helped spark a flame of interest in language AI when computer vision was the dominant branch of deep learning.

In the following years, Gomez saw how transformers took over the field and influenced the whole NLP landscape at Google. But he also noticed that most companies without the bottomless budget of Google couldn’t deploy the huge multi-billion parameter models that were emerging as state-of-the-art (SOTA) in NLP. It’s then when he saw the opportunity to create CohereAI: A company intended to bring down barriers and become “the default NLP toolkit” for developers.

I had the opportunity to talk with Aidan Gomez, CEO at Cohere, about its mission, vision, and goals. Here’s why Cohere is worth keeping an eye on.

Cohere in a nutshell

Gomez, who co-founded Cohere two years ago, says the main attraction of the company is its…

--

--

Alberto Romero
Alberto Romero

Written by Alberto Romero

AI & Tech | Weekly AI Newsletter: https://thealgorithmicbridge.substack.com/ | Contact: alber.romgar at gmail dot com

Responses (1)