Oh hi! New Website, split in two, welcome (if you’re a subscriber reading this in your inbox we have 2 new websites, you should check them out).
Quick cheatsheet, because anything I’ve tried to write about these spins in circles and makes less sense in my head than when I started, so I can’t imagine seeing it from the outside for the first time.
Orange:
Localeaf
The Nursery and Gardens.
Back to our family’s roots, our own personal learning and rewilding journey, and our adventures and misadventures in becoming a nursery. This year, opening to the public!
Hot pink:
Canopée Localeaf
our new NFP, encompassing all our community projects, learning programs, workshops and events, as well as the amazing group of volunteers, Pollinate Aylmer, who help create connected pockets of habitat across Aylmer.
You can switch between them through the colour bars at the top of pages on either site.
Both of these websites were made possible by this incredible person who walks through life holding my elbow when I stumble, and has yet to fail to help me build something insanely impractical that I dream up. I challenged him to his limit this time, and he spent months from the side of his desk building out the jumbled box of tangled cables of ideas and fragments of ideas in my head.
Check out both sites. They are pretty cool. Check out our Canopée Aylmer Map. Look what we’ve accomplished so far. See what we’re planning next. Join Pollinate Aylmer. Help us do more.
Our new Plant Database
On the new nursery website, everything pales in comparison to our new Plant Database. It’s currently in Beta, and encompasses every plant we’ve sown seeds for this year, as well as several years worth of personal research through academic sources. It started this season as a 54 column spreadsheet of things I liked to know about plants I grow.
I convinced a Principal Software Architect at Microsoft to help me, and realized just how much he loves me in the process, because we got through the past few months of trying to programmatically and systematically untangle what was tangled in my brain, and he still makes me the world’s best coffee and kisses me good morning.
How we made it
We aggregate data from a huge bibliography of government and peer reviewed sources, grouped by authoritative subject matter and tiers of credibility. Quite a lot of data. You can see it all here.
This data includes detailed plant morphology, which is later fed through an image generation engine that creates the botanical drawings and cards you see. Why? I love block printing, I love historical botanical drawings, and I wanted to begin with a scientific depiction of plants. I’ve also played around with AI image generation tools and found them to be very unstable and unpredictable, but realized that if we have a model create data-driven images, they would essentially become mathematically sound and as predictable as the quality of data the rendering is based on. And the images are SOOOOO pretty.
This is not a random AI generated database. All of the data used is programmatically retrieved. The fields that are calculated by LLM have been programmed to do so with the constraints and rigour used in scientific research. Every detail displayed about a plant has a traceable research basis. The LLM fields (computed or deduced by LLM based on criteria that has been tested and adjusted for months to reach this point) are clearly identified in our bibliography. If you are still with me and wondering how we were able to do this, you would have to understand the power of Colin’s brain, and knowledge, and brain again. I like to think I had a bit to do with this, and I recognize that his gigantic brain stepped in to compensate for where mine misfires, but our brains have worked together and built some pretty crazy things in the past, and it’s been lovely to see we still can.
What’s next?
I owe my partner in crime a thank you, and felt inspired to dig up this beauty from early 2019.
Thank you for being my unicorn
Our next planned steps are to address any remaining data interpretation anomalies first, particularly around location pinpoints, then decide what additional fields are missing that we would like to expose, as we already have the data (eg. Germination and seed care, Deer and rabbit resistance, etc. )
Eventually, we plan to pull in every plant historically native to the Outaouais and Gatineau area. We’ve reached a point where our model can do this, but need some help working out the remaining kinks. If you are a data driven geek like us, know plants, and want to help, please reach out. We would love the help.
Leave a Reply