No Bullshit AI Cannabis Course
Use AI to run your cannabis business better. No hype, no jargon.
Short lessons that show you exactly what to type, what to expect back, and when AI will waste your time. Written for the people running dispensaries, grows, labs and distros, by a software team that talks to them every day.
Short lessons on models, prompts, automations, agents and connectors, each scoped to a real operator problem: Metrc mismatches, invoice matching, menu data, compliance paperwork. Read this level with the person on the other side of the table.
Lessons on LLMs, prompts, agents, MCP servers and n8n workflows, scoped to real problems: Metrc reconciliation, invoice matching, menu data, compliance paperwork. Every lesson ships a usable artifact and works with whatever you already run.
Start where you like
Every page here reads at three levels, from plain English to full jargon. Move the dial at the top any time, or answer a few questions and I'll pick a starting point for you.
A few quick questions, one tap each. No wrong answers, and nothing is saved anywhere but your own browser.
You were reading at Plain English last time, so that is where I have put you.
Skipped the level check. The dial at the top of every page does the same job.
Buddy puts you at Plain English 1
You can move the dial at the top of any page whenever you like. Nothing here is saved anywhere but your own browser.
The courseThe coursemodules/
- 00
Start HereStart Here (how the dial works)Orientation
How the course works, why every page reads at three levels, how the sections and exams fit together, and a tour of every interactive part.How the site works: one page, three reading levels, five sections with an exam each, and one of every interactive part to try.Three-level rendering, the section and exam structure, the exercise kit and the playground, and how to contribute a lesson.
- 01
AI Basics, Minus the HypeAI Basics: LLMs and Decision Models, Without the HypeFundamentals: LLMs and decision models
Two kinds of AI, side by side. The kind that writes (ChatGPT, Claude) and the newer kind that only decides: pick one, rate it, yes or no. What each gets wrong, what you can safely paste in, and a first real task with each.Two kinds of model: LLMs that generate text, and System One decision models that return typed answers with calibrated probabilities. Failure modes, data-retention, and a first useful task for each.Tokens, context windows, sampling and failure modes for LLMs; state, typed questions, calibrated probabilities and confidence for System One decision models; data-retention policies; a zero-to-useful setup for an operator's first workflow with each.
- 02
How It Actually WorksHow It Actually Works: learning, words as numbers, attentionUnder the Hood: learning, embeddings, transformers, sampling
Why the tool behaves the way it does: how a machine learns from examples, why it sometimes memorises instead, how it turns words into numbers, and why the same question gets different answers.How machine learning actually works (training from examples, overfitting), how words become numbers (embeddings), how a transformer keeps track of a sentence (attention), why answers vary (sampling and temperature), and the different kinds of model.Symbolic AI to ML to LLMs, perceptrons and gradient descent, overfitting and generalisation, embeddings and cosine similarity, attention and transformers, sampling and temperature, the model taxonomy.
- 03
Pick the Right ToolPick the Right Tool: Model, Automation or AgentModel & Architecture Selection
Which AI for which job, when you need a chat versus an automation versus a little robot, when a decision model beats a writing model, and how to find the thing someone already built.Which model tier for which job, when a fixed automation beats an AI, when a decision model (pick / score / yes-no) beats an LLM, when you actually need an agent, and how to find what someone already built.Model selection by task, LLM vs. System One decision model vs. deterministic automation vs. agent loop, search-first sourcing, and token economics.
- 04
Prompts That Actually WorkPrompts That Work for OperatorsPrompting for Operators
One prompt structure, how to hand it your spreadsheets, how to get emails and SOPs out of it, and the habit of checking its work.One prompt structure, how to hand it your spreadsheets (CSV in context), getting emails and SOPs out, and checking its work.A single operator prompt template, tabular data in context, document generation with style anchors, and verification patterns.
- 05
Cannabis Problems, Solved With AICannabis Problems: Metrc, Inventory, Sales, 280EDomain Workflows: Compliance, Inventory, Sales, Accounting
Real problems from real operators: Metrc mismatches, inventory exports, buyer spreadsheets, 280E bookkeeping, and messy menus.Real operator problems, each solved with the smallest tool that works: a spreadsheet diff, an API call, one model step, or nothing at all.Problems pulled from Distru support tickets, each solved with the smallest tool that works: CSV diff, API call, one LLM step, or nothing at all.
- 06
Automations Without a DeveloperAutomations Without a Developer (n8n, CSV)Automation Patterns
CSV round-trips, an n8n workflow in an afternoon, deliveries on your calendar, and what to do when it breaks.CSV round-trips, an n8n workflow in an afternoon, calendar and Slack notifications, and what to do when an automation breaks.CSV bulk edits with validation, n8n schedule/HTTP patterns against the Distru API, calendar sync, and reliability basics.
- 07
Skills, SKILL.md, and MCP ExplainedSkills (SKILL.md) and Connectors (MCP)Skills, SKILL.md, and MCP
A skill is a text file that teaches AI how you do something. MCP is the phone line that lets AI use your software. Neither is complicated.What a skill file is, what MCP connectors are and what they cost every turn, then build a compliance skill and wire up one connector.SKILL.md anatomy and triggers, Model Context Protocol servers and tools, building a compliance skill, and wiring an MCP server into your tools.
- 08
Using AI With DistruUsing AI With Distru (API, MCP, agents)Distru API, MCP, and Agents
For Distru customers and people considering it: get your data out, ask it questions, turn buyer emails into draft orders, and build a small status-update robot.The Distru API in 20 minutes, asking your inventory questions from an export or a read-only connector, AI order intake, and a status-update agent with guardrails.Distru Public API v1, CSV-in-context versus a read-only tool layer, AI order intake, and an agent for order state transitions with guardrails.
- 09
MoonshotsMoonshots: Genetics, Your Own Model, a BenchmarkResearch Directions
The big ideas we are not sure about yet: AI helping design strains, a cannabis-trained model, and a test to find out which AI is best at this industry.Where AI in cannabis might go: strain genetics, a domain-tuned model, and a public benchmark. Honest about what is unproven.Genomic models for cultivars, fine-tuning vs. retrieval for domain models, and a public cannabis benchmark. Honest about what is unproven.
- 10
Building It YourselfBuilding It Yourself: the Developer TrackImplementation: the call, the loop, the bill, the proof
For whoever writes the code. What a request actually is, how a tool loop works, how to stop it costing more than it should, and how to know it still works next month. Read this at level 1 if you want to understand what your tech person is doing.The developer track. One request end to end, the tool-call loop and where the write gate belongs, caching as an architectural choice rather than a setting, and an eval built from your own transcripts.Messages requests, the tool-use loop and parallel results, prompt-prefix design and measured token spend, and a private eval with rubric grading. Four lessons, written for the person with the API key.
- 11
Your Own DocumentsYour Own Documents: search by meaning and answers from your filesRetrieval: semantic search and RAG over your documents
Stop pasting the right page into the chat by hand. Search your SOP binder in your own words, get answers that point to the page they came from, and see what it takes to build one.Search by meaning over your own documents, retrieval-augmented generation (RAG) that answers from them with citations, and a small working build you can adapt.Embedding indexes and semantic search, the RAG pipeline (chunk, embed, retrieve, generate with citations), its failure modes and evaluation, and a stdlib reference implementation.
- 12
Ship It SafelyShip It Safely: responsible AI, trustworthy features, and keeping them workingOperating AI: responsible deployment, trust-centred UX, LLMOps
Before an AI feature goes in front of staff or customers: the ways it can hurt someone in a regulated business and how to stop them, how to design it so people can check and correct it, and how to keep it working after launch.Responsible AI for a cannabis business (measure, mitigate, operate), designing AI features people trust (sources, confidence, control, feedback), and the lifecycle after launch: evaluation, monitoring, model updates and cost.Harm identification and layered mitigation, trust-centred interaction design, and the LLMOps loop: evaluation gates, monitoring, pinned models, incident response and governance.