We build AI tools
for conservation
Conservation organisations are not short of data. They are short of the hours and the engineers to use it without taking time away from the conservation work itself. We train the models that read it, and the hours go back into the field.
Why we exist
Conservation technology keeps getting built twice. Budgets are thin, so the work goes to whoever will take it at that price, and the technical decisions get made with nobody in the room able to judge them. Two years on, the person who understood the system has moved to another job, the hosting bill has lapsed, and four seasons of data sit on a drive in an office while somebody starts the whole thing again.
We think that is a licensing and maintenance problem rather than a technology problem. So everything Ceder builds is released under the Apache 2.0 licence, and written to be picked up by someone who did not write it. What we release stays released. Nobody can withdraw it or start charging you for it later, including us.
We are named after the Clanwilliam cedar, which grows on about 250 square kilometres of these mountains and nowhere else on earth. It has not come back on its own. Where it is coming back, it is because people carried seedlings up by hand and climbed back years later to see which ones took. Slow work, done in the open, by people who planned to still be there.
Find the animals in your data
A season of camera traps comes back as tens of thousands of images, most of them grass moving in the wind. We find the frames with an animal in them and name the species, then do the harder part: matching the same leopard across every camera she walked past, this season and last. That is what turns a pile of photographs into a number you can put in a report.
Listen to a whole landscape at once
Sound reaches where a camera cannot see. Through fog, after dark, across a valley, into bush too thick to walk. A recorder left out for a month comes back with several hundred hours of it, nearly all wind and insects and engine noise.
We tune the open acoustic models to your species list and your own background noise, so what comes back is every call with a time and a recorder against it, and the clip behind each one so your team can check the ones that matter. From that you can say which sites a species is still using, and whether that has changed over time.
How you get it
Everything we build is open, ready to be used and adapted to your needs. If you have someone technical, take it and run it yourself. There is nothing to sign and nothing to pay.
If you do not, we do the engineering with you. You are not starting from zero: everything the organisations before you funded is already in there, built and working. What you pay for is the part that does not exist yet, built around your conditions, and it joins the library the same way.
Tell us what you are working on
And what is not working about it at the moment. We will tell you whether it is something we can help with, and what it would take. If the honest answer is that you should take the open tools and run them yourself, we will say that instead.
Get in touch