Section 01, The case for the post-harvest decade.

Twenty-plus years in cannabis. Ownership in ten facilities. Toured 500+ grows across four continents. Equipment built at Keirton has processed over eighty million pounds of cannabis to date. From that vantage point, one pattern is hard to miss: the industry has spent a decade optimizing cultivation, and very little of that energy has reached post-harvest.

Genetics, lighting, nutrients, environmental control, plant care, that's where the money and talent have gone. Post-harvest has been treated as a labor problem to be managed rather than an engineering problem to be solved. It's the stage most operators measure least precisely, and the stage where yield, quality, and labor are actually won or lost.

$1.4M / yr, a five-point yield difference on five thousand pounds per month at greenhouse pricing. Most operators cannot tell you their post-harvest yield within five points. They can tell you plant counts, cycle times, HVAC tonnage, and nutrient costs to two decimals. The gap between cultivation rigor and post-harvest rigor is the single largest unforced error in cannabis economics today.

This report is a framework for closing that gap. It maps the architectures producers actually use, what each one costs, what each one yields, and where AI is now changing what's possible.

Section 02, The five architectures of post-harvest, and where the dollars go.

Commercial post-harvest is five stages: drying, bucking, trimming, sorting, and curing. This report focuses on the middle three. Within those, work can be done by hand, by machine, or in a hybrid arrangement. The five common configurations, with cost and yield ranges seen at scale:

• Full hand processing, best yield (70%–low 80s), highest cost ($35+/lb).
• Typical hybrid (machine + QC), $20–$30/lb, 45%–70% yield. Cost and yield are inversely coupled via QC labor.
• Lean hybrid, lower cost, lower yield. Lever is yield, not cost.
• Full automation, $10/lb floor, lower yield, higher B-grade percentage. Appropriate for pre-roll-dominant operations.
• Slow full hand, high cost, mid yield. Lever is labor speed.

Quality is not a third axis on this table; it is a constraint. Push cost down or yield up far enough and the flower is no longer worth what the price column assumes.

The framework anchors on greenhouse-grown flower at 2,000–10,000 lb/month. The architectures translate to indoor and outdoor; what changes is wholesale price and labor profile.

Section 03, Three things break the cost-yield tradeoff. AI is the mechanism behind all of them.

Labor reduction. AI vision systems grade flower faster and more consistently than humans. In one production deployment, AI sorting replaced a fifteen-person line for $750,000 in annual labor savings.

Yield improvement. AI between the trimmer and QC pulls flower out of the machine stream early when ready, and routes the rest for the right intervention. Across three independent production-scale evaluations on greenhouse and outdoor flower, this approach delivered yield gains of 3.6% to 8%. At greenhouse pricing on 5,000 lb/month, a 5-point yield gain is roughly $1.4M/year.

Intelligent stream separation. Defect screening removed ~3% of pre-pack flower as mold-affected, protecting product quality and reducing recall risk. Value-tier sorting recovered 37% of "smalls" as medium-or-larger and routed them to a higher-tier brand at a $150/lb price uplift.

Across five independent producer evaluations, AI-driven systems delivered documented annual value of approximately $1M to $6M per producer. AI vision typically adds roughly $1–$2 per pound to the operations cost basis at production scale.

Section 04, Where post-harvest goes from here.

Three forces are driving post-harvest into its decade: economic (dollars per pound at stake are large and growing), technical (AI vision is now commercially deployable in production), and competitive (margin advantages compound during consolidation). AI will move from single-stage tasks to managing whole automation lines as integrated systems, generating data that feeds back into cultivation, genetics, and operational planning.

Find where you sit. Identify the lever that matters for your operation. Execute. The producers who do this work in the next two to three years will define what best-in-class looks like for the rest of the decade.