From Growing in my basement . . . .

I accidentally started building an AI agriculture company because I was trying not to kill plants in my basement

Hey Indie Hackers,

I’m Ryan, founder of CanopIQ.

This whole thing started in a pretty normal, slightly ridiculous way: I was trying to grow in my basement and I took a picture of one of my plants because I could tell something was off, but I didn’t know exactly what.

I put the picture through AI just to see what would happen.

At first, it was mostly curiosity. I wasn’t trying to build a company. I was trying to answer the very scientific question of, “Is this plant healthy or am I slowly murdering it with confidence?”

The AI response was not perfect, but it was interesting enough that it stuck with me. It could see things I was seeing, but not always naming. Leaf color. Curling. Possible nutrient issues. Environment problems. Stress signals. It made me realize that most growers are already collecting tons of information, but it is scattered everywhere.

Photos on phones.
Sensor dashboards.
Nutrient notes.
Grow journals.
Spreadsheets.
Memory.
Random text messages.
The one guy in the building who “just knows.”

And in controlled-environment agriculture, especially cannabis, that scattered information matters. A small issue can turn into a bad crop cycle. A bad crop cycle can become real money, real waste, and real operational pain.

My background is healthcare IT, so my brain immediately went somewhere maybe a little different than a typical grow app. I’ve spent a lot of time around regulated workflows, documentation, clinical systems, pharmacy technology, audit trails, and messy operational software. In healthcare, you do not just say “patient looks bad.” You track history, context, orders, interventions, outcomes, and accountability.

So I started thinking:

Why don’t plants have a real record?

Not just a journal.
Not just a sensor graph.
Not just a camera feed.
Not just a chatbot answer.

An actual plant-response record.

That idea became CanopIQ.

The first version was rough. Very rough. Like “this might be brilliant or I need to go outside and touch grass” rough.

But I kept building.

The core idea is simple: CanopIQ helps controlled-environment growers understand how plants respond over time. It uses plant photos, grow observations, environmental context, facility notes, and operational records to help detect issues earlier, document what happened, and turn grower experience into something more repeatable.

Cannabis is the first wedge because it is high-value, regulated, and unforgiving. But the larger vision is controlled-environment agriculture in general: cannabis, food crops, research grows, education, resilient local production, and eventually any grow where the plant response matters more than just the room temperature.

One of the things I learned early is that growers do not need another dashboard just for the sake of having a dashboard. They need better visibility. They need earlier warning signs. They need better documentation. They need to know whether a plant is still on track, not just whether it is still alive.

That became one of the core CanopIQ beliefs:

Most systems monitor the room, control equipment, or track operations. CanopIQ interprets plant response across the whole grow stack.

I built the first versions with Gemini, and Gemini has honestly been a huge part of how fast I was able to move. But I do not think of CanopIQ as a Gemini wrapper. The product is the workflow, the plant record, the context layer, the scoring, the history, the reporting, and the way everything connects back to grow decisions.

The AI model is an engine. Important, yes. But swappable.

If privacy, cost, customer requirements, or compliance make it necessary, I can move the AI layer in-house using local or private LLMs and vision models. That matters to me because I do not want the whole company to depend on one outside model provider forever. I want CanopIQ to be model-flexible, not model-trapped.

Over the last several months, this has gone from “I wonder what AI thinks of this plant picture” to a real product direction.

Right now, CanopIQ includes things like plant profiles, AI plant analysis, health scoring, encounters, grow timelines, tasks, grow journals, product/input tracking, and a professional dashboard. I’ve been building toward a version that can support both small growers and commercial operators, with a path into more serious controlled-environment workflows.

I also started putting myself and the company into rooms where people would be forced to either validate the idea or politely destroy it.

I participated in Techstars Founder Catalyst, which helped me sharpen the company, the story, and the path forward. I don’t want to oversell that as some magic stamp of success, because it is not. But it did help me take CanopIQ more seriously as a company instead of just another weird Ryan project living in twelve browser tabs and a basement.

I’ve also been working with Dr. Paul Rushton, a plant scientist with deep experience in plant biotechnology, cannabis/hemp research, and grant strategy. That has been a big deal for me because CanopIQ needs to be more than “AI says your leaf looks sad.” The science matters. The plant biology matters. The way we frame this for agriculture, research, food security, and grants matters.

We’ve been looking at grant opportunities and broader applications beyond cannabis too. Cannabis is the first serious commercial wedge, but I think there is a bigger story here around controlled-environment agriculture, resilient food systems, education, and helping more people grow successfully with better tools.

I’m still early. I’m not pretending this is a finished empire.

I have a working product.
I have demo workflows.
I have grower conversations.
I have grant work in motion.
I have a basement demo site coming together.
I have a much clearer idea of what the company is than I did when I started.

And I have a lot still to figure out.

The biggest challenge right now is turning the product from “technically working and interesting” into something growers will consistently pay for and use. That means better onboarding, tighter workflows, clearer ROI, more real-world grow data, and probably a lot of uncomfortable conversations with people who will tell me what is actually useful versus what is just cool.

That last part is the hard part. Cool is easy. Useful is where the rent gets paid.

I’m posting here because I want to start building this more publicly and connect with other founders who are building in weird, niche, operationally messy spaces.

I’m especially interested in talking to:

• people building AI products that are not just chatbots

• founders working in agriculture, cannabis, compliance, sensors, or field operations

• anyone who has sold software into old-school or relationship-heavy industries

• anyone who has gone from “I built this myself” to “real customers are using this and paying for it”

• growers or operators willing to be blunt about what would actually help them

CanopIQ started because I was standing in my basement, staring at a plant, wondering what I was missing.

Now I’m trying to build the system I wish I had then: something that helps growers see earlier, document better, and make smarter decisions before problems become expensive.

That’s the backstory.

Now I’m trying to see if I can turn it into a real company.