Highlights:

  • The platform provided by Replicate ensures to alleviate the manual tasks of software teams.
  • The company provides a collection of thousands of open-source AI models, as a part of its feature set.

A startup, Replica Inc. announced raising USD 17.8 million funding and exited stealth mode. This is going to make creation of Artificial Intelligence (AI) applications simpler for software personnel.

The company received these funds over two rounds. It raised USD 12.5 million, a bulk of the capital, from Series A investment that involved Sequoia Capital, Y Combinator, Andreessen Horowitz, and several other angel investors. Earlier, the company successfully closed a seed round with a deal worth USD 5.3 million.

It is often technically tedious to integrate AI models into an application. Generally, neural networks can’t execute by themselves, rather require a deployment with some auxiliary code segments that are usually challenging to write and coordinate.

The deployment of AI-backed applications on graphics card is another complex part of the process. Sometimes, integrating graphics card in an application environment complicates the task of infrastructure management.

The platform provided by Replicate ensures to alleviate the manual tasks of software teams. The startup claims that the platform helps developers to deploy AI models with merely a few lines of code.

Firshman, Chief Executive Officer, reported “There are roughly two orders of magnitude more software engineers than there are machine learning engineers (~30 million vs. ~500,000). By building good tools, we think it is possible for software engineers to use machine learning in the same way they can use normal software.”

The company provides a collection of thousands of open-source AI models, as a part of its feature set. As AI models perform a wide range of tasks from text translation to image generation, its platform also functions with custom neural networks developed in-house by the company.

Neural networks are generally depoloyed with auxiliary tools or components such as TensorFlow and PyTorch that help accelerate AI development. In certain cases, developers also incorporate advanced components such as CUDA toolkit by Nvidia Corp, which is used to enhance AI models’ performance deployed on Nvidia chips.

Developers are often expected to assemble the model and supporting components into a software container prior to the AI model deployment. This process involves writing Dockerfile, a well-known code file. It specifies the components and their configuration to be included in the container.

Replicate has developed Cog, an open-source tool that soothes the task. This tool in the company’s platform allows users to customize containers with fewer configurations than the task usually takes. A Dockerfile generated by the Cog can specify the configuration of container according to the limited set of instructions coming from software teams.

The platform simplifies the task of deploying AI models in production. After the neural network is packaged into a container by developers, Replicate automatically deploys it to the managed cloud environment. Further, the environment can add and discard hardware resources as graphics cards according to the usage.

Replicate is becoming immensely popular in the developer ecosystem. Since mid-2021, user base of the startup is expanding at a month-over-month growth rate of 149%, according to a report from a renowned online media.