Smart SOD · Soil Organism Detector

AI biodiversity monitoring

Reading the living soil, one organism at a time.

We pair large-bore flow cytometry with machine learning to identify the invertebrates, microfauna, and microbes that make soil alive — at an unprecedented resolution and scale.

Explore the California ATBI How Smart SOD works

Scroll

Soil samples processed

Organisms extracted

01 — What we do

Smart SOD workflow: a soil core is suspended and passed through the cytometer's custom laser suite for multispectral imaging, then classified into a soil-organism record. — Smart SOD — sample to classified record

A soil microbiome is a census no one has finished taking. We built the instrument to take it.

SoilTech is a biodiversity-monitoring company. Conventional soil surveys rely on hand-sorting, morphological identification, and weeks of specialist labor — capturing a fraction of what lives in a single gram of soil. Our platform extracts, sorts, counts, saves, and identifies the live organisms in minutes.

Whole community

Nematodes, Collembola, Tardigrades, rotifers, and more…

High throughput

Soil samples in minutes — over 2,000 organisms per minute.

Preserved specimens

Organisms preserved for rearing, barcoding, and future study.

02 — The technology

How the Smart Soil Organism Detector works

The Smart SOD is a custom large-bore flow cytometer paired with a machine-learning pipeline — combining high-resolution imaging, multispectral sensing, and flow cytometry to extract, count, identify, and separate intact soil organisms, non-destructively and reproducibly.

A soil sample tube combined with a water buffer to form a suspension.

STEP 01

Extract & suspend

A soil sample is dispersed into a buffered suspension that keeps whole organisms — roughly 5 to 500 µm — intact for analysis.

Organisms flowing single-file through the custom laser suite, scattering green and red beams.

STEP 02

Multispectral cytometry

Organisms pass single-file as four solid-state lasers (405, 488, 561, 640 nm) scan each one at 10 MHz across 12 optical channels, with a 5 MP brightfield camera imaging at 5×.

A multispectral digital fingerprint — a heatmap of optical-channel signals over time for one organism.

STEP 03

Digital fingerprint & ML

Each organism becomes a multispectral digital fingerprint, and a trained model assigns a taxon with a confidence score — up to 95.5% accuracy identifying nematode strains.

Sorted specimens deposited into the wells of a plate for preservation.

STEP 04

Count, separate & record

Counts and identities are recorded non-destructively, and organisms can be physically separated and preserved for rearing, barcoding, and future study.

Peer-reviewed research

The Smart Soil Organism Detector

Filgueiras, Kim, Wickings, El Borai, Duncan & Willett

Biosensors and Bioelectronics 221 (2023) 114417

View at publisher (DOI) ↗

Digital phenotyping accelerates soil biodiversity discovery

Filgueiras, Kim, Gluesenkamp & Willett

Manuscript

Download PDF ↗

03 — Data & access

Every record, openly published and ready to use

Open the data portal

FileFormat

Flow cytometry laser scans

.bxr4

Raw high-resolution imagery

.dcimg

Processed imagery

.png

Per-organism summary statistics

.csv

Summary information

.csv

More information is in the data portal.

DOWNLOAD

Start with a per-sample summary CSV, or pull the full archive — raw laser scans, processed imagery, and per-organism statistics.

DIRECT ACCESS

Pull the entire dataset straight from Amazon S3 (US East 1) at high speed — bulk transfers, no pagination or rate limits, requester-pays. Browse the inventory first to see what's there.

aws s3 sync s3://soiltech-cal-atbi/ . --request-payer requester

LICENSE

CC BY 4.0 — free to use with attribution. Full provenance on every record.

04 — Featured findings