Save time realizing AI visions

Ambolt AI’s new developer tool, Emily, makes configuration and deployment faster and simpler, allowing you to spend more time on the AI project itself.

A combination of tiresome configuration and a lack of technical resources keeps many companies from taking advantage of AI. With this in mind, Danish development house Ambolt AI has built its own tool to solve the challenge.

Their basis for doing so was, among other things, the fact that business leaders from almost all types of industries see enormous potential in AI. That, and the fact that many are worried about whether they are running fast enough – or whether the competitors are even faster at reaping the benefits.

This was, for example, one of the key findings of a leading KPMG report from 2021 entitled Thriving in an AI World. The report also found that small and medium-sized companies in particular are experiencing significant difficulty in choosing the right technologies for their AI initiatives.

Ambolt AI recognizes both findings.

Often a need for something other than standardised AI platforms

“Providers such as Microsoft, Google and IBM have exciting and sophisticated AI platforms that can solve standardised tasks efficiently. But the platforms are a poor match if your needs are not strictly standard, and that holds many companies back,” explains Jon Andersen, CCO of Ambolt AI.

GDPR or the company’s own privacy concerns also prevent some from using cloud-based AI platforms, forcing them to build their own. But this is easier said than done if you ask developer Anders Brams.

At Ambolt, we’ve experienced first-hand how difficult it is in practice to set up the underlying infrastructure to allow you to develop and deploy AI. You can spend an eternity configuring the development environment, allocating resources and tinkering with deployment. And if two developers each have their own OS or driver version, you can easily spend days just getting that under control.
Anders Brams

Emily makes light work of correct configuration

Download and test Emily in your own environment

But you won’t even get this far without full-stack developers in the organization with knowledge of how the elements are configured correctly, and in the words of Jon Andersen “they certainly don’t grow on trees”.

Ambolt AI therefore built Emily, a tool which can assist machine learning engineers and data scientists in their work with the development and deployment of ML microservices.

“Emily is a CLI tool that keeps track of the entire project from project start to deployment in a production environment. Emily moves the developer’s code editor into a container to ensure a fully containerized development environment. Similarly, all writing of code, testing and implementation of the microservice takes place inside the Docker container which will later be put into production. If your project works in the development environment, it will also work in operation,” explains Anders Brams.

“And because the development environment is fully containerized, set-up conflicts and hassle with different OS and driver versions are also a thing of the past,” he adds.

Emily is built up around a question framework that guides the user safely all the way from the first steps in the configuration to deployment – and the tool writes the underlying Python code along the way, ensuring that the configuration runs correctly.

How: Take the first easy steps towards effective AI development

Over the past year, the tool has made time-consuming and often frustrating configuration tasks significantly easier and faster for Ambolt’s own developers and data scientists. Jon Andersen recently demonstrated it to a number of customers, who were blown away by the potential.

“The idea then arose to make Emily available to others, too, so they could see how it could help them realize their AI projects,” he says.

The tool is available in two versions, both of which can be downloaded here – namely a basic version called Emily Core, which is available free of charge, and the brand-new Emily Deploy, which has deployment functionality.

We have enjoyed Emily ourselves, and I really hope that it can help others handle a large part of the tedious and often insurmountable tasks that make implementing AI plans and ideas so difficult.
Jon Andersen