Case Study

Digital Marketing Through Machine Learning

Content Engine was founded with the goal of bringing years of experience in digital marketing to businesses looking for a more efficient and affordable way to do content marketing. While traditional content agencies use manual labor to produce content, Content Engine produces original content through machine learning. This can significantly reduce delivery time and allows for scaling to any level of demand. Thousands of articles can be produced, at high quality, in a much shorter time than it would take a traditional agency.

• Digital marketing company using machine learning to help companies generate highly-engaging content
• Uses TensorFlow and PyTorch and a range of Natural Language Processing frameworks
• Currently working with two government clients and hoping to have commercial services available to the public.

Prior to using Spell, Content Engine was using custom Notebooks to work on Natural Language Processing experiments, but didn’t have a well ordered approach for tracking and organizing experiments. Training their machine learning models was also potentially very expensive. As a result, they decided to use Spell to move their model training to AWS and GCP, which allowed them to scale up to large compute requirements and develop a better organized approach to running experiments. 

Natural language models oftentimes require very large amounts of memory and compute power.

Upon adopting the Spell platform to run their Natural Language Processing models, Content Engine was able to use their existing AWS and GCP credits, helping offset the potentially high expense of training their large models. 

The team at Content Engine found Spell to be a very straightforward and streamlined approach to getting started. With an easy-to-use UI side and CLI, Spell was ideal for creating new workspaces, runs, and interacting with the integrated filesystem. 

Tri Ngyuyen

Partner and Chief Product Officer

For moving to larger scale environments, Content Engine says Spell is the only option for them that truly works. Some of their models have massive memory and compute requirements, and moving those to the cloud was the only practical way for them to run experiments. Spell enabled them to seamlessly move their heavy NLP models to the production and commercialization phase. 

To learn more about Content Engine, visit contentengine.ai, where you can also receive a free content sample and join their pilot program.

Schedule an in-depth demonstration with a Spell representative to learn how Spell can help streamline and accelerate your machine learning development.