About Gilhe

The high-speed format and quality transformer. Gilhe is a small London engineering studio that obsesses over one number: median time to first download. Most of our British users arrive after a frustrating round trip with a social platform that turned a sharp photograph into mush, so the way we measure ourselves is the wall-clock between “upload finished” and “download link clickable”.

The number that drives the studio

The internal target is 500 milliseconds at the median, on commodity hosting, including server-side decode, encode and HTTP delivery. Below that figure the user perceives the operation as instant. Above it, support tickets begin to rise. Every feature we ship passes through that gate; if it nudges the median above 500 ms, it gets refactored or rolled back. The discipline is why the home page contains exactly three controls and no signup wall.

Built around social re-encoding

Most online converters tune for benchmark images. Gilhe tunes for what happens after your file leaves us. Default JPEG quality 80 and WebP quality 85 are the figures that survive Facebook, Instagram and LinkedIn re-encoding with the least visible degradation. Those numbers came out of three years of measurement on real campaigns from a half-dozen UK marketing teams. They are not arbitrary.

Why we keep the tool free

Gilhe is supported entirely by reputable display advertising. We have not added a paid “pro” tier because the existing free tier already covers the work most users actually have. The conversion-tool category is full of opaque tools that quietly retain uploads or hide quality controls behind a paywall; we want to demonstrate that the basic version of the product is genuinely sustainable as the public version, not as a teaser.

Privacy as an architecture choice

Working files are wiped automatically within twenty-four hours by a sweeper that runs on every request. We do not keep originals, hashes or thumbnails for analytics. The cropper runs entirely inside your browser tab, so cropped photos never reach a Gilhe server in the first place. We do not train models on uploaded image content. If your organisation needs a Data Processing Agreement under the UK GDPR for higher-volume use, write to the address below.

Editorial line

The articles on this blog are written by our in-house engineering team. We do not accept payment to feature specific brands, products or formats. We test every recommendation on a real workload before publication, and we revisit articles when browser support, encoder behaviour or platform spec sheets change. If you spot something out of date, the contact e-mail goes to a real person.

Getting in touch

For partnership requests, advertising enquiries or press, please use the same e-mail address with a subject line that starts with “Press” or “Partnership”. We aim to reply within two business days.

Meet the editor