From a day to an hour: choosing a CMS for an AI company

SKY ENGINE AI builds synthetic data for training computer vision models. Content covered long technical articles, product explainers, landing pages, newsletters, social posts and sales materials, and much of it needed review by data scientists and other experts.

Before the CMS, content lived on the CEO’s computer and was hard-coded into the website’s design. Any change needed the front-end team.

Publishing a piece took about a day.

Even small text changes were difficult, and publishing something new was harder still – every update had to go through the front-end team.

1. Built a set of requirements for the CMS, based on both our content needs and the front-end team’s needs. We wanted a headless CMS and full control over what was published, when, and on which pages – with features such as a draft stage, scheduled publishing and API connections.
2. Found a number of candidates and evaluated each one against every requirement.
3. Narrowed the field to three contenders and recommended my first and second choice to the C-suite for approval. They approved my first choice: Hygraph.
4. Took part in moving the company website to the new CMS: I helped migrate the old content and reviewed or wrote the UX copy for the new site.

~1 day → ~1 hour

Publishing time cut by around 90%.

I might have chosen a good-enough provider instead of trying so hard to future-proof our content operations. Hygraph came with a steep learning curve, and getting through it was uncomfortable.