Spring planting season is well underway, and by now your growers' data is flowing. While they were prepping planters and planning for planting, the Leaf team was making sure their data would flow faster, from more equipment, and at the highest quality possible. We shipped over 1,400 updates across our API and Dashboard. Here are the highlights.
New Equipment Integrations
CNHI FieldOps
Leaf now supports connections to CNHi's new FieldOps platform, which runs alongside the existing AFS Connect. FieldOps is CNHi's next-generation platform and API, giving you access to field boundaries and field operations data. Note that FieldOps is a separate platform from AFS Connect and unfortunately existing AFS Connect credentials won't carry over. This means your growers will need to establish new connections to your applications through FieldOps. Once connected, their data flows through Leaf automatically and will be standardized just like existing data from AFS Connect. See the CNHI FieldOps documentation for setup details.
Precision Planting Panorama
Precision Planting Panorama is a new OEM provider on the Leaf platform. If your growers use Panorama, you can now connect their accounts and get field data and operations synced, converted, and standardized through the same pipeline as every other manufacturer. As with other OEMs, you need to set up your developer accounts and API agreements with Precision Planting in order to connect through Leaf. See the Panorama documentation for setup details.
Leaf Lake: SQL Access to All Your Data
This is the biggest new capability we've shipped since GeoParquet. Leaf Lake gives you direct SQL access to all the machine and fields data. Instead of downloading files and processing them on your end, you send a SQL query to a single endpoint and get back JSON rows. USDA soil data (SSURGO) is also available. It works with standard SQL aggregations, CTEs, JOINs, and spatial functions.
Contact us to get access to Leaf Lake.
Faster Operations
We've made substantial improvements to how quickly operations are created and updated. We rebuilt the merge process, which combines individual machine files into field operations, on a new architecture, and the results have been great: operations are created roughly 2x faster on average compared to six months ago. Your growers' data gets from raw file to queryable operation faster than ever. This is just the beginning. Our team has some even more ambitious work in development that will speed file processing up by at least an order (or two) of magnitude.
Take Control of Your Data Filtering with cleanupRules
Previously, cleanupStandardGeojson was a binary toggle. Either all default filtering rules applied, or none of them did. Now you have full control.
With cleanupRules, you define exactly which properties to filter and what thresholds to use. Each rule specifies a property (like harvestMoisture, seedRate, or appliedRate) and a condition using operators: greater than, less than, equal, not equal, and their inclusive variants.
For example, you could require harvest moisture greater than 5% to filter out zero-rate application points, or drop points where crop equals "unknown" — all while keeping the rest of the default rules. When you set custom rules, they replace the entire default rule set, so you know exactly what's being applied.
These same rules are also available in Leaf Lake, so you can experiment with different filtering criteria at query time without reprocessing anything.
See the full reference in the configuration documentation.
File Conversion & Processing
We want your files to just work. Regardless of which equipment produced them, our team is constantly improving how we parse, convert, and standardize machine data. Here's what changed this period:
New format support
- Kinze KML: Leaf now maps planted properties from Kinze KML files
- Raven JDP: a new Raven format variant is now processed through the pipeline
Data quality improvements
- GPS calibration: a new processing step corrects GPS drift patterns in machine data. If your growers' field passes don't line up quite right, this can help. Enable it with calibrateGpsPoints — contact us if you'd like to try it.
- Elevation filtering: invalid z-coordinates (elevation spikes from GPS glitches) are now dropped automatically
- Heading calculations: improved for multi-section equipment where each section may report slightly different values
- Distance and area: recalculated more accurately for point-following equipment
Crop and variety
- New crop name mappings handle non-standard naming from various monitors
- Better handling of files with mixed timestamp formats in the crop mapping flow
Implement width
Leaf now includes implementWidth in the standardGeojson and standardGeoparquet output. This tells you the working width of the implement and is useful for calculating coverage, overlap, and application efficiency.
GeoJSON is now optional
Parquet is always generated. If you only consume Parquet (and many of you do), you can now set generateGeojson to false and skip GeoJSON generation entirely. This speeds up processing, especially for large files.
Converter improvements
- John Deere: products and implement width now extracted from JDL files
- CNHi: improved CN1 binary conversion reliability
- ISOXML: updated ISO plugin for better Task Controller compatibility
- DAT/Climate FieldView: improved DAT conversion services
New Output Formats: DAT and ADM
While we think GeoParquet is ultimately the way to go (sorry GeoJSON), we recognize there are even more formats our industry relies on.
DAT Generation
Need to get data back into Climate FieldView DAT format? Leaf can now generate DAT files from your processed data. New API endpoints let you retrieve the generated files and check their processing status. Contact us to enable this.
ADM Generation
Similarly, Leaf can now produce ADAPT ADM files from processed data. This is useful if your downstream systems expect ADAPT-compatible files. Contact us to enable this.
More Pipeline Transparency
Understand what happened to your files
- Data collection files: these now get an "ignored" status instead of failing, making it clear they were intentionally skipped rather than broken.
- Partial data on failed files: when a file fails processing, the response now includes whatever
standardGeojson data was extracted before the failure.
- Products in output: product information is now included directly in the leaf file output and operation summaries
Leaf Dashboard Updates
The Leaf Dashboard is evolving into the primary interface for managing your Leaf account and visualizing machine data. While your services and agents might prefer our API, we know that human users still value a clean UI. Our team has been busy shipping substantial improvements to the dashboard experience over the past few months.
Better maps
- Stepped maps: custom color-coded visualization of field data properties, so you can see yield or application rate distributions at a glance
- Big Maps: gigabyte sized files that previously couldn't render in the browser now load reliably
- GeoParquet: maps now load from GeoParquet when available, which is significantly faster than GeoJSON for large operations
Billing visibility
You can now see your data processing usage directly in the dashboard with per-user breakdowns, sparkline trend charts, and granularity filters so you can drill into daily, weekly, or monthly usage.
Provider management
Magic Link improvements
- Magic links for provider connection are back on the grower detail page
- New
/fields step in the magic link flow lets growers choose which specific fields they want to enable for data processing so you're not paying to process fields your growers don't care about
Grower management
- You can now edit grower profiles directly in the dashboard
Organization Management
Dashboard access is now possible with more granular access control. This is especially useful for larger teams where you want to separate access by role or region. Organization management is available in the new Leaf Dashboard at dashboard.withleaf.io.
| Config |
What it does |
| cleanupRules |
Define exactly which data points to keep or filter, per property |
| generateGeojson |
Skip GeoJSON generation when you only need Parquet |
| calibrateGpsPoints |
Correct GPS drift patterns during processing |
Looking Ahead
With planting data coming in strong, we're already working on the next round of improvements. As always, we want to hear from you: what would make your integration easier, your product better, or your support calls shorter?
Drop us a line on Slack or reach out to your customer success manager. And if you want to try Leaf Lake, enable DAT/ADM generation, or turn on GPS calibration, just ask. Most of these are a quick configuration change on our end.
