The dahlias died last year. Every single one of them, spectacularly and without apology, despite my best efforts and not checking one single YouTube tutorial. My garden is small and deliberately manageable. I have historically approached it with the same energy I bring to anything I don’t fully understand - curious, slightly chaotic, and unreasonably optimistic about outcomes.

I know plants are pretty but also necessary. I know some of them smell extraordinary and roses without a beautiful heady fragrance are hardly deserving of the name. Something quietly satisfying happens when a thing you planted actually survives winter and bursts into life come spring. Beyond that, I am largely making it up as I go.

Which is why I didn’t see this one coming.

For every Field Notes edition, I ask a writer to share something from their speciality. I consider it their turn at the microphone - their domain, their depth of knowledge, their turn to teach me what I need to know. They do not get told what to write.

When I approached Mike D, MrComputerScience, while I didn’t quite know what he would decide, I was fairly confident of the direction – he’s our AI news guy, right?! We read his carefully curated round-up of the events that have shaped the previous week and appreciate the way he breaks each news story down, showing us why it matters and how it affects us personally. Mike has an incredible skill for teaching and I asked him to bring that over as a guest of Daring Next.

I was not expecting the direction he chose.

It absolutely delighted me because I had no idea what to do with it when Mike first introduced his plan.

We spend a lot of time in the AI space drawing lines. The domains AI suits, the work it enhances, the things it shouldn’t touch. Lines feel safer than the alternative - admitting the map keeps changing faster than we can draw it.

Nature was supposed to sit on the other side of one of those lines. Ancient, analogue, sensory. The domain furthest from anything a token prediction system should reasonably touch.

Mike has been getting extraordinary results from AI for three years, in a domain nobody expects - permaculture, native ecology, backyard conservation - and the reason isn’t that he’s technical. It’s that he’s precise. His approach works because it's built around his exact location, his specific ecosystem, his particular patch of ground.

Generic question, generic answer. Specific question, specific plan. There is something worth noticing in where Mike chose to direct that precision. The same technology the world is debating for its data centre footprint is, in Mike's backyard, figuring out which native plants will save the pollinators.

The ask is everything. Including the question of what’s worth asking for.

For me, this week was a good illustration of that, though not from a garden. My last newsletter was a real-time account of what happens when something that was working just stops working in a dramatic way. Claude and I had been off for weeks, the collaboration sour in a way I couldn’t immediately diagnose. This week, I pulled apart every Skill document, every project instruction, every accumulated rule, and rebuilt from the ground up. I got Claude to help design its own instructions in ways it would actually use, removed the conflicts, cleared out what had calcified into dead weight. The result: Claude sings again (hoorah!). Boundaries are being held. The responsiveness is back. The tool changed. What I handed it changed too. The second part was mine to fix.

Which is, it turns out, exactly what my guest this week has been doing in his backyard for awhile now.

Not with AI instruction documents but with ecosystems. With pollinators and native soil and the particular plants that belong to that specific corner of the living world. The prompts Mike brings to this newsletter aren’t remarkable because they’re about nature. They’re remarkable because they’re built from the inside out — his location, his soil, his ecosystem.

The skill of describing your situation precisely enough so that another intelligence - human or artificial - can actually serve you, is not a technical skill. It is the skill of someone who spent years writing emails to get busy people to do things they could not be bothered doing - who learned to lead with the why, not just the what, who built context into every request, who explained the impact, created conditions where another person could actually help, and who got results through relationship and back-and-forth rather than hoping a single perfect sentence would do the work.

That is context engineering. You have probably been doing it your whole working life. You just haven’t always seen it operating when you’re standing in a garden, or at a desk at half past eight, or in a conversation with an AI that keeps giving you answers that feel like they were written for someone else.

Mike has been living what he calls a double life. A deep expertise in permaculture and native ecology on one side, AI collaboration on the other. Kept mostly separate because gardeners, like many communities built around something that feels sacred, have complicated feelings about the technology. He’s decided it’s time to stop keeping them separate. He’s brought three prompts: practical, hyperlocal, designed to help you grow food, support pollinators, and take action for at-risk species in your specific backyard.

You don’t need to be a gardener. You don’t need to love AI. Just be curious enough to try one prompt and see what happens.

The dahlias are staying out of it this time. I’m taking notes.

Share

Welcome, Mike D!

Mike D is an AI journalist, editor, and educator from Greater Boston with over 37,000 total enrolments on Udemy. Author of Pithy Cyborg | AI News Made Simple. Roots native plants and neural networks with equal enthusiasm.

How to use AI to grow food, support pollinators, and protect endangered species

Most people think AI is pulling us away from nature.

I’ve found the opposite.

For the last three years, I’ve been using AI to plan gardens, identify native plants, support at-risk pollinators, and design food systems. My yard has never been more alive. My harvests have never been better.

And I’ve never felt more connected to the ecosystem around me.

But here’s what I haven’t told nearly anyone, until now. I’ve been living a double life.

I’ve edited a gardening newsletter for years, where I’ve published hundreds of articles on permaculture, food forests, and native plant ecology. I’m obsessed with soil microbes, pollinator guilds, and succession planting calendars.

But many gardeners don’t trust AI. Some actively hate it. They see it as the opposite of what they value. Synthetic, extractive, disconnected from the earth.

That’s the problem. The narrative that “AI versus nature” is a binary choice is lazy. And it’s keeping people from tools that could genuinely help them live more self-reliant, ecologically informed lives.

You can use AI to identify which crops will give you superior yields, which native shrubs will have the highest impact on pollinators, and how to support at-risk wildlife in your backyard before they disappear forever.

Some say AI kills nature. I say AI is the lens through which we finally see the geometry of the wild.

That’s why Dallas and I are sharing three AI prompts designed to help you reconnect with the land. These aren’t theoretical exercises. They’re practical, hyperlocal plans you can deploy today to grow food, support wildlife, and live closer to nature.

If you’ve ever wanted to:

• Plan a spring garden, but didn’t know where to start

• Support pollinators, but didn’t know which plants are actually native

• Help at-risk species, but didn’t know which ones live near you

These prompts will give you answers far superior to generic advice, Instagram aesthetics, or garden guru hype. Instead, you get plans tailored to your climate, your soil, your ecosystem.

You don’t need to love AI. You don’t even need to trust it.

You just need to be curious enough to try one prompt and see what happens.

Sound fair?

Here’s how to use AI to see your backyard through a thousand seasons of insight.

Prompt 1: Use AI to brainstorm your ideal spring garden

Every successful garden begins with one variable that trumps aesthetics, enthusiasm, or Instagram inspiration: your growing zone. Climate determines planting windows, frost risk, crop viability, and long-term resilience. Get this wrong, and you’re fighting nature. Get it right, and nature does half the work for you.

The prompt below generates a precise, location-specific planting plan.

Instructions:

Your location is the only input required. Look for the text labelled “MY LOCATION” near the top. Then, enter your location and paste the entire prompt into Perplexity, Claude, GPT, Gemini, or any modern AI chatbot of your choice.

The Prompt:

You are an expert in permaculture gardening and a local horticulture specialist. Please create a 
personalized spring garden plan based on my location and growing conditions.

MY LOCATION = [City, State, ZIP code, or Country/City/Town/Location.]

Based on my location, please provide:

SECTION 1: GROWING ZONE ANALYSIS

1. Geographic Anchor: (Based on the user's location.)
2. Hardiness Classification: Identify the USDA Hardiness Zone if the location is within the 
United States. For international locations, identify the equivalent RHS (UK), EGR (Europe), or 
localized minimum temperature bracket.
3. Thermal Boundaries: Average Last and First Frost Dates.
4. Temporal Window: Total Growing Season Length (Annual Frost-Free Days).

SECTION 2: TOP 15 CROPS FOR MY ZONE

For each crop, provide:

1. Crop name
2. Seed starting date (indoor if applicable)
3. Transplant/direct sow date (outdoor)
4. Expected harvest window
5. Space required (square footage or plant spacing)
6. Growing method (seed vs transplant recommendation)
7. Maintenance level (Low/Medium/High)
8. One elite growing tip (the thing most beginners miss)

Focus on:

1. Mix of vegetables, herbs, and pollinator-friendly plants
2. Beginner-friendly crops with high success rates
3. Crops that provide continuous harvest through succession planting

SECTION 3: COMPANION PLANTING GUIDE

Create 3-5 "plant guilds" that reveal which crops grow well together and why (pest control, 
nutrient sharing, space optimization).

SECTION 4: SUCCESSION PLANTING CALENDAR

Show when to plant second rounds of fast-growing crops (lettuce, radishes, beans) for continuous 
harvest through fall.

SECTION 5: COMMON MISTAKES TO AVOID

List 5 mistakes beginners make in my specific zone and how to avoid them. 

Make this actionable, specific to my climate, and designed for someone who wants to succeed on 
their first try.

✅ Why This Prompt Works:

This is Precision Agriculture scaled for the individual.

It replaces generic templates with Site-Specific Intelligence and transforms static data into a living roadmap. It also encourages Symbiotic Mapping (Companion Planting) and Temporal Sequencing (Succession Planting).

This AI prompt literally shifts your garden from a single harvest to a Recurring Yield Engine.

Prompt 2: Use AI to find the best native pollinator plants for your growing zone

Gardening isn’t only for you. Done right, it’s life-saving infrastructure for essential pollinators, insects, and birds. The issue is that most “pollinator-friendly” plants sold at big-box nurseries aren’t native. Native plants evolved with local species, support exponentially more life, and are far easier to grow. The prompt below identifies the native pollinator plants in your region.

Instructions:

Your location is the only input required. Look for the text labelled “MY LOCATION” near the top. Then, enter your location and paste the entire prompt into Perplexity, Claude, GPT, Gemini, or any modern AI chatbot of your choice.

The Prompt:

You are a native plant ecologist and pollinator conservation specialist with deep knowledge of 
regional ecosystems across North America and the world. I need you to create a personalized 
native pollinator garden plan for my specific location.

MY LOCATION = [City, State, ZIP code, or Country/City/Town/Location.]

Based on my location, please provide:

SECTION 1: ECOSYSTEM CONTEXT

1. My ecoregion
2. Native soil type common in my area
3. Key native pollinators in my region
4. At-risk pollinator species I can specifically support

SECTION 2: TOP 12-15 NATIVE POLLINATOR PLANTS

For each plant, provide:

1. Common name & Latin name
2. Plant type
3. Bloom period
4. Height & spread
5. Sun requirements
6. Moisture needs
7. Primary pollinators attracted
8. Wildlife bonus
9. One growing note

Prioritize:

1. Plants that bloom at different times
2. Mix of flower shapes
3. Native plants readily available at local nurseries
4. Perennials over annuals

SECTION 3: SEASONAL BLOOM CALENDAR

Timeline showing which plants bloom when for continuous nectar flow.

SECTION 4: PLANTING STRATEGY

1. Best planting season
2. Spacing recommendations
3. Soil preparation tips
4. Where to source plants

SECTION 5: ECOLOGICAL DESIGN TIPS

1. Arrangement for maximum pollinator traffic
2. Companion plants
3. Leave seed heads and stems standing over winter

SECTION 6: WHAT TO AVOID

List 3-5 invasive ornamental plants in my region.

✅ Why This Prompt Works:

This is Ecological Succession Engineering.

Most “pollinator gardens” fail because they only provide nectar windows. But not nectar highways.

This prompt ensures Temporal Continuity (bloom overlap across seasons) and Functional Diversity (flower morphologies matched to pollinator guilds). The “what to avoid” section acts as an Invasive Species Filter, preventing well-intentioned ecological sabotage.

You’re architecting a 200-day nectar corridor that supports specialist pollinators that most gardens starve.

Prompt 3: Use AI to identify at-risk species in your region and how to support them

Conservation often feels abstract. ‘Save the bees’ sounds important. But what does it actually mean in your backyard? This prompt turns vague concern into targeted action. Because planting natives is a good start. But if you want to make a measurable difference, you need to know which species are actually struggling and what they need from you. The prompt below surfaces which at-risk species live in your region and translates that data into concrete actions.

Instructions:

Your location is the only input required. Look for the text labelled “MY LOCATION” near the top. Then, enter your location and paste the entire prompt into Perplexity, Claude, GPT, Gemini, or any modern AI chatbot of your choice.

The Prompt:

You are a conservation biologist and habitat restoration specialist with expertise in regional 
biodiversity and threatened species management. I need you to create a personalized action plan 
for supporting at-risk species in my specific location through backyard habitat creation.

MY LOCATION = [User inserts: City, State, or ZIP code]

Based on my location, please provide:

SECTION 1: LOCAL ECOLOGICAL CONTEXT

1. My ecoregion and dominant historical habitat type

2. Primary threats to local biodiversity

3. Key ecological relationships that have been disrupted

SECTION 2: AT-RISK SPECIES IN MY REGION

Identify 5-7 at-risk or declining species I can realistically support in a backyard setting.

SECTION 3: TARGETED HABITAT ACTIONS

For each at-risk species, provide 2-3 concrete actions.

SECTION 4: HOST PLANTS & SPECIALIST RELATIONSHIPS

Identify 5-8 native plants that serve as irreplaceable host plants.

SECTION 5: NESTING & OVERWINTERING HABITAT

Guidance on creating habitat structure for bees, birds, and beneficial insects.

SECTION 6: WHAT TO STOP DOING

List 3-5 common backyard practices that harm at-risk species.

SECTION 7: MONITORING & IMPACT

Suggest 2-3 citizen science projects or apps to track impact.

✅ Why This Prompt Works:

This is Conservation Biology scaled to the backyard.

The Host Plant Matrix (Section 4) targets Obligate Relationships, meaning the non-negotiable dependencies that determine whether a species can complete its lifecycle in your region. Many gardens function as Ecological Traps. They attract adults but provide no reproductive substrate.

This prompt transforms your yard from a pollinator restaurant into a pollinator nursery. The “What to Stop Doing” section is Harm Reduction Ecology. It eliminates the invisible killers (mowing schedules, fall cleanup, pesticides) that negate planting efforts.

You’re now running a species rescue operation with measurable conservation impact.

Try one. Share what happens.

I don’t need you to love AI. I don’t need you to trust it completely. Pick the one that solves a problem you actually have. Run the prompt. See what it generates. Then go outside and do the work.

Because after three years of using AI to reconnect with nature, I’ve learned something that most Instagram “garden gurus” won’t tell you. The tool doesn’t replace the dirt under your fingernails. It just gets you to the right dirt faster.

• AI won’t plant the milkweed. You will.

• AI won’t build the food forest. You will.

• AI won’t notice the first monarch caterpillar, or hear the Carolina Wren nesting, or taste the first pawpaw from the tree you planted five years ago.

But AI will tell you which milkweed is native, which shrubs provide nesting cover, and which fruit trees will actually produce in your climate.

If even one person uses these prompts to help native pollinator species or supports an at-risk butterfly, this was worth it.

Cordially yours,

Mike D (aka @MrComputerScience)

P.S. If you have gardening questions, ask. I’ve been doing this long enough to have made every mistake at least twice. If you try one of these prompts and it changes how you see your backyard, that would genuinely make my year.

Mike ends with a line worth sitting with for a moment: AI won’t plant the milkweed. He will. You will.

The tool doesn’t do the thing. It gets you to the right thing faster and closes the distance between not knowing where to start and knowing exactly which plant belongs in which corner of your specific piece of ground. Mike ran his prompts and got a hyperlocal, ecosystem-specific, seasonally calibrated plan. He still had to go outside. He still had to get dirt under his fingernails. What changed was that he arrived at the right starting point rather than a generic one.

That’s the thread running underneath everything here. Underneath Mike’s garden, underneath my rebuilt Claude instructions, underneath every piece of work you’ve ever done where the output actually matched what you needed. The people who get mediocre results from AI are asking it for advice that could apply to anything. The ones who get genuinely useful, specific, actionable responses are describing the situation precisely, building the context, making it possible for another intelligence to actually help.

You already know how to do this. You’ve been doing it for years in every process you made legible for someone who couldn’t see the shape of it yet, in every conversation where you translated the messy reality of a situation into something another person could actually act on. The garden is just the most unexpected place it’s shown up this week.

The dahlias are getting another shot this year. I have considerably more specific questions than last time, and considerably less faith in optimism alone as a growing strategy.