Ever wondered if your greenhouse control system is actually helping—or if it’s leaving you in the dark about your energy bills? I’ve been there, and that’s exactly what we dig into in this episode.
Joining me are Neda, CEO and co-founder of Microclimates, and Gretchen, Executive Director at the Greenhouse Lighting and Systems Engineering Consortium at Cornell, whose combined expertise covers everything from cutting-edge environmental controls to energy efficiency in controlled environment agriculture. Neda has an extensive background in developing technologies that empower growers with actionable data, while Gretchen brings years of research experience with leading universities and utility-backed initiatives in optimizing greenhouse lighting and automation.
This episode unpacks the real-world findings from the CalNEX Project—a first-of-its-kind scientific study focused on the impact of smart environmental monitoring and controls in California greenhouses. We compare “smart” and “smarter” systems, revealing surprising industry gaps in environmental data collection, misunderstood overhead costs, and how simple steps can lead to significant energy savings and business sustainability.
Beyond the results, we chat about practical steps for adopting automation without ripping out your current systems, new open-platform sensor trends, the reality (and future) of AI in controlled ag, and why a phased, data-driven approach will be key for small and large growers alike. If you’re daunted by all the talk of sensors, integration, or AI, consider this the guide to understanding what actually matters—and what you can do today.
Curious if you’re missing an easy win in your farm’s energy management, or want a reality check on all the AI hype? Tune in now and turn your environmental data into your biggest asset!
Thanks to Our Sponsors
CEA Summit East - https://indoor.ag/cea-summit-east-2025/
Indoor AgCon - https://indoor.ag/
Key Takeaways
00:00 Discussing the CalNEX Project findings
03:17 Smart Controls Project achievements
07:52 Improving greenhouse energy efficiency
10:34 Energy consumption misconceptions
12:57 Diverse systems in cannabis farming
19:10 Automating greenhouse light management
22:22 Operational impacts in construction projects
25:40 Consulting on greenhouse tech needs
27:50 Discussing phased approach for innovations
29:57 Phased approach to AI implementation
35:51 Parental influence on IT security
37:02 Adopting AI and Automation Tools
41:17 Starting with basic crop monitoring
43:32 Indoor farming control techniques
46:39 Year of retrofits and opportunities
49:46 Appreciating industry partnerships
Tweetable Quotes
"Our job is to complement what's already there. The projects Gretchen mentioned, every site had something different… Our job wasn't to go in and just say, rip everything out for this control study and start over because we want to collect the data. It was really to take a look at what do you currently have, how can we complement that with adding more environmental insight, environmental visibility to that operation, and is there a way that we can integrate what you currently have?"
"I'd say the biggest surprise and biggest lesson was that they didn't have enough environmental visibility. A lot of these operations had maybe one temperature humidity sensor hanging in the middle of the room representing the entire greenhouse or a section of the greenhouse. Your control system is only as good as the information it's taking in, like the input, right?"
"If you don't have the thousands of data points, if you can't summarize the trends, if you can't make any recommendations and alerts, and you can't generate those reports based on the thousand data points, how is this AI going to actually get to know you?... First of all, we have to step back and say, are operators actually collecting data? We already said early on in this conversation they are, but they don't have enough environmental visibility, which means they actually are not collecting enough data."
Resources Mentioned
Website - www.microclimates.com
YouTube - https://www.youtube.com/@Microclimates-Inc
Instagram - https://www.instagram.com/microclimates
Facebook - https://www.facebook.com/microclimates.inc
LinkedIn - https://www.linkedin.com/company/microclimates
LinkedIn - https://www.linkedin.com/in/gschimelpfenig/
Resource Innovation Institute - https://resourceinnovation.org/
Microclimates - https://microclimates.com/
Priva Control Systems - https://www.priva.com/
Ritter Greenhouse Automation - https://rittergreenhouse.com/
Connect With Us
VFP LinkedIn - https://www.linkedin.com/company/verticalfarmingpodcast
VFP Twitter - https://twitter.com/VerticalFarmPod
VFP Instagram - https://www.instagram.com/direct/inbox/
VFP Facebook - https://www.facebook.com/VerticalFarmPod
Subscribe to our newsletters!
AgTech Digest - https://agtechdigest.com
The Indoor Farmer - https://www.indoorverticalfarm.com/
Horti-Gen Insights - https://www.hortigeninsights.com/
🎙️🎙️🎙️
Podcast Production and Marketing by FullCast
Discover the best podcast services in the world at The Podosphere: https://www.thepodosphere.com/
Mentioned in this episode:
2025 Precision Ag Report by iGrowNews
1
So Gretchen and Neta, no strangers to the Vertical Farming Podcast. Thank you
2
so much to both of you for joining. I know I've been in conversations
3
with Neta about an opportunity to come back on the show, and she mentioned some
4
work you guys were doing together. So just to kick things off, Neta, you want
5
to kind of talk through what prompted the desire to come back
6
on and share what you've been working on? Yeah, absolutely. Thanks again for having us
7
on the show. It's good to always chat with you again, Harry. It's always,
8
we love chatting with you because it's this organic conversation where you
9
really highlight what's going on in the industry. And what really prompted me to come
10
back and have a conversation was to shine some light
11
on a project that we worked with ERI. Gretchen
12
brought us into a project which was called the CalNEX Project, which was
13
really taking a look at the energy consumption that's being used in the
14
greenhouse on operations and really understanding
15
does adding more environmental monitoring and controls actually
16
make a difference from an energy usage perspective? This was
17
the first study, from what Gretchen described to me, that was really
18
being looked at from a very scientific perspective. And having had a science background,
19
we were really excited to be a part of this project, which is really looking
20
at things side by side, kind of what we call the smart room
21
versus a smarter room over a long time period. So really exciting to
22
be back on this show to talk to you guys about what the findings were
23
and what Gretchen's team and was able to learn from this project.
24
Thanks for that context. And Gretchen, I think what would be helpful for the viewers
25
is to kind of share how long you've been involved in horticultural lighting research
26
as some context leading into this collaboration. Yeah,
27
sure. So in 2021, I was on a team
28
that worked with one of the largest utilities in North America, Commonwealth
29
Edison, that serves like Chicago and a lot of areas in
30
Illinois. To explore what they called the opportunities in controlled environment
31
agriculture. And so at that time, I started this theme of looking into
32
and uncovering the benefits of LED lighting and automating those
33
systems. And not just the energy benefits. At that time, I was actually looking at
34
what the utilities call non-energy benefits. So I think that's what growers
35
often care about more, right? Is, yeah, energy savings, but also what
36
else? What happens to the plants? What happens to the the labor benefits and other
37
things like that. So then in 2023, I took on the part-time role of being
38
executive director of GLAZE, the Greenhouse Lighting and Systems
39
Engineering Consortium at Cornell University. And I got to collaborate with amazing
40
scientists, still get to collaborate with amazing scientists at Cornell
41
University, Rutgers University, and Rensselaer Polytechnic Institute.
42
And at that time, I was helping them complete this NYSERDA-funded
43
greenhouse lighting research. Which took a look at not just the beginning
44
of LED lighting adoption, but also the adoption of
45
dynamic lighting controls. And that's where we're going as I get to the
46
point of where I started to work with Netta. So in 2023,
47
you know, I had been running into Netta in the industry quite a bit, and
48
we collaborated a bit when I was at Resource Innovation Institute. But in
49
2023, now at Energy Resources Integration, I applied for funding
50
like we all try to do. We try to get some funding from someone else
51
to do something cool. So I applied for funding from a California utility program
52
called CalNEX. And what they do is they explore energy efficiency
53
technologies and vet them for inclusion in rebate
54
programs, right? So how can we get new tech to get money so
55
that growers can adopt it, so that any business can adopt it? So from
56
2023 to 2025, Neta and I collaborated on the Smart Controls
57
Project, where we explored how these dynamic
58
controls for lighting, climate control, as well as energy monitoring, like Netta mentioned,
59
really prove out the energy benefits so the utilities would give out rebates,
60
but also prove out the benefits for the businesses so that they would want to
61
do it too. So we completed 4 field demonstrations at 3 different
62
farms in California. And, you know, this year we're really running
63
around celebrating the results, which is that it proves that these
64
will beneficial and that utilities should provide rebates. And hopefully
65
the wave will just be beginning now of rebate programs
66
offering more for controls, and more research and education
67
about this will continue to see growers adopt it more. So that's my history for
68
the past 5 years in this field, and I really loved the past few years
69
being able to collaborate with NETA. Sounds like you guys had a— it's like a
70
match made in heaven in terms of what NETA's been working on Neda, what made
71
Microclimates a good fit to support this project with CalNext?
72
Yeah, I'd say the biggest thing was probably that we are
73
unique in the sense that we're not trying to replace any environmental
74
control system. Our job is to complement what's already
75
there. And the projects that— the locations that
76
Gretchen mentioned, every site had something different. One of them had
77
a Ritter control system, a very robust, great company. Another
78
one had a Priva. Another one didn't really have much
79
automation. So, our job wasn't to go in there and just say, rip everything out
80
for this control study and start over because we want to collect the
81
data. It was really to take a look at what do you currently have, how
82
can we complement that with adding more environmental insight,
83
environmental visibility to that operation, and is there a way that we can integrate
84
what you currently have? Will that control system allow you to integrate
85
And what does that look like? So it really just sort of demonstrated how additional
86
environmental monitoring and not ripping and replacing can really bring a lot
87
of value. And what did that experience teach you about what is
88
being collected now? Was there anything surprising when you went in?
89
Yeah, I'd say the biggest surprise wasn't necessarily that they
90
weren't collecting data because they were all collecting data. It was really
91
the biggest surprise and biggest lesson was that they didn't have enough environmental
92
visibility. So, what I mean by that is that a lot of these
93
operations had maybe one temperature humidity sensor hanging
94
in the middle of the room representing the entire greenhouse or a section of the
95
greenhouse. And your control system is only as
96
good as the information is taken in, like the input, right? So, it really shined
97
a light on the fact that they didn't have enough environmental
98
visibility. And the perfect example that I love referring to
99
is Floricultura, one of the sites.
100
Their crop is an orchid. So they're growing orchids, very
101
finicky plant, as we all know. Yeah, very finicky. But they really just didn't really
102
understand the environmental temperature, humidity
103
environment that the crop was actually experiencing at a root zone
104
level, below the root, above the crop, above
105
their screens. So that was really important. And we have a grower there
106
that's very data-driven, data-rich. He really
107
understands orchids. He's probably world-renowned for his understanding of orchids.
108
And having that insight was so valuable to him. And Gretchen
109
can talk a little bit more about what the outcome of that study was, but
110
that was a kind of aha moment that we had, was, yes, they're collecting
111
data, but they don't have enough crop environmental
112
visibility. And then the question that really came for us at the end was,
113
if they're not collecting enough data as it is, How's this world going
114
to prepare them for AI, which we can talk about later? But that really got
115
us thinking from an AI perspective as well. So yeah, she's teed that
116
up for you perfectly, Gretchen, the experience with that. And I'm also curious how you
117
pull in all these learnings that you have from all these other locations
118
you've been at, all this other research you've been doing on smart controls. And I'm
119
curious what your perspective was specifically with that company that Neda
120
mentioned. And then what comes to mind for me is like, Does the then grower
121
have to think about, oh, I need more sensors now to capture all these different
122
locations and all these different places? And I'm sure there's concerns there as well. Yeah,
123
I think the— I'll reiterate Neta's surprising moment as
124
well of realizing that the field demonstrations are
125
using pretty static controls or pretty basic
126
controls. So that presented an opportunity for us. I
127
think that in the floricultura example, They have a sophisticated
128
control system, but what they don't have is a lot of feedback loops
129
telling them what's happening at the grower level. The system knows, and it
130
adjusts the screens, and it might adjust the lighting, but they aren't perhaps getting
131
the info where they can find fault detection. Another site at the
132
lettuce greenhouse, the load shape of the lighting circuit looked pretty much the same
133
every day, even as the seasons were changing. And so my job as an engineer
134
when I'm validating a technology for a utility is, what's the
135
baseline? What would they do if we didn't affect anything, if we didn't
136
offer a rebate? And so for me, this helped us prove
137
that scheduling lighting with time clocks is an industry standard
138
practice. And so utility energy efficiency programs
139
could help growers save even more energy by using sensor-based controls. And
140
instead of just using schedules, they could use sensors to control their lights. With
141
floriculture, they're not using lights as much, but with lettuce, the lighting
142
was much more of an important thing. So with mushrooms, one of our other
143
field demonstrations, the humidity and the climate values were much
144
more important. So it's sometimes what we found was surprising
145
because it was like, there's a great energy savings opportunity. And then sometimes what we
146
found was more surprising of like, wow, there's a real big like information
147
gap here that Netta's system can help fill by just providing
148
things like trends. Because, you know, I'm used to commercial buildings
149
where you can go back, open up the computer log, see months of
150
trends, see what all the office building temperatures were in all the different
151
rooms. But some growers we found, all they knew from a
152
climate level was something that was at a very, like Netta said, one sensor
153
in one place. And then all they knew from an energy level was the utility
154
bill. For the whole greenhouse. So, Nada was helping us piece apart,
155
okay, well, this is actually how much is going to lighting, and this is what
156
the lighting's doing. This is how much is going to fans, and pumps, and all
157
the other stuff that sometimes is more energy-consuming for a different type of grower that
158
doesn't use lights. So, Nada, I imagine a lot of this information
159
you've been learning in your interactions with your clients, and the folks you've been
160
working with, and all the facilities you've been able to have access to. But, how
161
much of this was a learning for you, and because of this partnership, because of
162
the in-depth visibility you had to what was there already,
163
and maybe some preconceived notions about what people think growers are
164
measuring versus what's actually happening in these facilities? It was
165
quite a surprise. I think I had the vision. I assumed that
166
they were collecting a lot of data, and then we, I think, also made some
167
assumptions that they understood a little bit about their energy consumption, at least
168
maybe like— because we, we've heard this over and over and over again the past,
169
you know, 10 years— overhead costs are too high?
170
What happened to the company that just started and they spent millions and raised
171
millions, hundreds of million dollars to build this beautiful vertical farm? And why
172
did they go out of business? Over and over again, the theme that we have
173
heard is that overhead costs, overhead costs. And we know that 70%
174
of their energy consumption is related to their HVAC
175
systems and their lighting system. So I think I had made the assumption that they
176
must understand something about their energy consumption. And even the companies
177
that had a whole sustainability team. We started out with one of them that had
178
a whole sustainability team dedicated to it. They really cared
179
about this topic, and they knew all about it, but they really
180
weren't collecting energy data at a circuit level,
181
at, let's say, a pump level, at a lighting
182
level. So, it was an aha moment of they actually don't
183
have visibility into that data, into how much energy is
184
being used. And if you don't have that visibility, How are you going to make
185
small changes so that you can reduce that overhead cost? Is that
186
common, Gretchen, from what you've seen? Obviously, you've done a lot of this research a
187
lot. And I'm also curious if there's a difference between controls that
188
need to be monitored on the greenhouse side versus pure vertical
189
farms, and if you have enough data to kind of back that up.
190
Yeah, I appreciate you bringing up that there's maybe different baselines for
191
greenhouses and vertical farms. And if you check out the results of my study
192
with Netta, If you look at what we did was we created a tiered
193
hierarchy of control sophistication levels. So level 0 is
194
basic. Level 0 is manual. Level 1
195
is when you start to have some basic controls like a
196
timer. And then when you start going to level 2, that's when we start to
197
have some sorts of an automation system going on, but they're not
198
talking to each other. And then as you go up the levels, you start to
199
have actual system integration. And what we tried to do with the study was for
200
every system, lighting, HVAC, and irrigation, what
201
level are they? So we did surveys, we did interviews, site visits,
202
and field demonstrations. And when you take a look at the levels of
203
all of those systems, you can't say that there's a common level that
204
you see amongst all those systems across greenhouses and vertical
205
farms. We observed cannabis indoor farms that
206
In some cases had manual lighting controls, which boggled my mind, but that's what
207
they chose to do. And then we also had one of the most, what they
208
touted to be the most sophisticated, completely integrated, completely
209
like well-oiled machine-controlled cannabis farm. We went
210
to a huge cannabis greenhouse, which had very little lighting and
211
therefore did not need a ton of advanced lighting control, but had one of the
212
most sophisticated irrigation systems we'd ever seen, where they did
213
gravimetric weighing of their plants to see how much water was happening for
214
a sample plant that therefore then dictated how much water the rest of the crop
215
got. They also had sophisticated energy generation
216
systems, cogeneration systems, and absolutely knew a ton about their
217
energy. So that's cannabis, and that's just how you can see, like, you've got the
218
full spectrum there. It's hard to say what the baseline is. Some people
219
have said that when you've seen one greenhouse, you've seen one greenhouse. So in my
220
study, I've seen many, but can we say that my study
221
applies to everyone? No, it probably just applies to California. But the last thing
222
I'll say is that I do think that overall, greenhouses, if they're growing
223
a crop that has a lower profit margin, we generally saw have less advanced
224
controls. And if you have a crop like we've talked about with
225
orchids, there's more of a reason to have some more Cadillac-level
226
integrated system. But I think what we do see overall across all of
227
them was still generally a fairly low level of system
228
energy monitoring. So they've got an idea of how much they're paying
229
for their bills, but they— most of them all across the board
230
did not have the ability to say, yep, I've been spending $20,000 on lighting,
231
$40,000 on HVAC this month, and then in total my bill was
232
$80,000. So yeah, that's sort of the gist of it, is like
233
the study helped us put more dots on the map, but we still have more
234
studies that still need to keep proving that
235
there's a general trend for a particular crop. I
236
can't really say that yet. So with all this data that you now have
237
available, Neta, I'm curious how you think about approaching
238
partners to work with. Does this change your thinking about maybe doing an
239
audit first about seeing where they're at and what exactly they're measuring? Because
240
to Gretchen's point, they may think they've got a robust monitoring system,
241
and with the proper audit, you can almost pick out like where
242
There are gaps in how they're thinking about this. And then obviously one of the
243
questions is going to be, is microclimates going to compete with the systems
244
or is it going to work or is it going to complement them? So I'm
245
sure those are questions that come up as well. Yeah. So I think what you
246
mentioned about the audit, yeah, I completely agree. Gretchen's team and
247
Gretchen's expertise and ERI is the perfect company
248
that can help with those things. They could step in and take a look at
249
an operation and audit them and understand Where, you know, you can even— I
250
believe, Gretchen, if I'm not wrong, you guys can even audit their energy bills and
251
see if there was some mistakes on their bill. So there's a whole level
252
that ERI and Gretchen's, especially Gretchen's expertise, can really
253
step in and help operators right away without having to install
254
anything. They could just look at documents and papers and bills
255
and make sense of it. That's number one. Then number
256
2, what you mentioned, Harriet, is Competing or
257
complementing? Well, historically, I'd have to say that most
258
companies out there have been about replacing, right? We actually come in and
259
complement. We're not asking to rip and replace. The other thing that's
260
really important is that these energy monitoring systems have to be easy
261
to install. If they're not easy install and there's going to be a bunch
262
of wires everywhere, it's unlikely that the operators want to go through the
263
hassle. And that's where I think what we've done at Microclimates is really
264
try to simplify that process. where we bring in a computer with our
265
EnvOS, which is an environmental operating system, and these wireless
266
sensors for energy monitoring that just connect, hook to the
267
circuit level, and quickly begin to monitor your environment. And
268
then the operator also has the ability to put in their scheduled
269
pricing. So then automatically the dashboard will show you
270
how much energy did this pump use How much did it cost
271
me? And you have all these beautiful graphs that you can look at and make
272
sense of it. Then you start working with a company again like ERI
273
and Gretchen's team, and you say, okay, now that you've reviewed my
274
baseline bills and what I'm doing, and now that
275
I've added a few circuits, I have 3 months of
276
data, 4 months of data. Maybe you want to have 1 year of data depending
277
on your crop, maybe seasonality, who knows. Now,
278
Gretchen, what do you suggest I do with this information? And that's where they
279
come in and help you fine-tune your system slowly. And again,
280
it doesn't have to cost very much. It's not like we're spending hundreds of
281
thousands of dollars by any means. It's very reasonable
282
to get to that point and shave off 5% off your energy
283
bill, which makes a huge difference. Yeah, Gretchen, I see you
284
nodding your head. So what's usually the responses when you show them these studies?
285
And obviously, there's always pushback when you have to change systems that are
286
already in place. And I'm curious, what are some reasons you'd give a greenhouse
287
operator for reasons why they should switch to
288
daylight-responsive lighting controls, for example? Yeah, or
289
daylight-responsive lighting controls plus some more energy
290
monitoring. Exactly. I think that the things that both
291
Nada and I try to do is take one step at a time. So yes,
292
There's small steps to take, like doing a bill audit or doing
293
a quick audit of like, what do you even control? What do you even
294
measure? And then I would be able to share with Neta, okay, they
295
have this much information. And what we can do together is provide them with
296
more valuable information on top of what they already have. And
297
then also provide the right thing, whether that's
298
light, the right light, or the right humidity, or the right temperature. Right.
299
And then lastly, the thing that I care about, save energy. So get
300
more information, get the right target hit, and save
301
energy. Hopefully save money because everyone's energy costs are going
302
up. And so the payback period just gets better the more you implement
303
automation. So we found that sometimes growers didn't know how much light
304
they were actually receiving, right? So they can then tailor their
305
controls to meet production goals, influence plant quality
306
or adjust the amount of light they're doing because they were never actually meeting DLI
307
target, which is something we've found in quite a few greenhouses we've worked with.
308
Then they can use, you know, automation to either
309
do some sort of simple algorithm or use AI, which we might talk about a
310
little more, to decide how much light should I give at any given, you know,
311
15-minute interval. Should I add a little more light because power
312
costs are going to go up tomorrow? Do we have a production goal we haven't
313
met yet? So there's some things that people can start to do to provide light
314
when it's needed or provide, you know, turning it off when it's not needed. And
315
then lastly, you know, the pocketbook. I think that we haven't talked about the numbers
316
yet, but I've been really excited because my study has sort of stood
317
alone along with some academic studies that have been done by like University of Georgia
318
and Cornell University for a while, since like 2016. These
319
academic studies plus the CalNEX study have said automation
320
saves a good amount of energy. And usually that raises
321
eyebrows and suspicion because large numbers usually mean like, how can you really trust
322
that you're gonna save 50% of my lighting energy? That's
323
crazy. But it's not actually crazy. My study, as well as those
324
recently performed by other teams, so the ASHRAE
325
Standard 90.1 committee just this past month presented
326
a slide that showcased 42 to 62% energy savings
327
compared to time clock controls when you implement daylight responsive
328
controls. So if you're paying over 10 cents a kilowatt hour, which
329
many of us are, many of us even at our house are paying more than
330
10 cents a kilowatt hour, the payback can be as short as months.
331
And I heard someone in the ASHRAE committee say this
332
may be the quickest payback measure that we've ever added to
333
90.1. And so I really think that speaks to not just
334
the impact, the magnitude of this opportunity that
335
for the CEA industry, but overall for the whole buildings industry,
336
this is a very big opportunity in terms of energy savings. So that's what
337
I'd say to a greenhouse owner, but also I'd say to a vertical farmer as
338
well. I'd say, what do you currently not know? What do you want to
339
hit for your target? Do you know you're hitting that target? And then let's save
340
some energy. Even if you're indoors, you're probably able to dim your lights
341
in ways that you're not. You're probably able to modulate your fans, your pumps.
342
There's always those tweaks, the continuous improvement that Netta was talking about.
343
5% a year, that'll help you cushion yourself from utility
344
cost increases. And then if you go even more aggressively, you know, if you have
345
a dedicated energy management practice, you could be saving more like 15%, 20%
346
a year. Or you could go big with one thing and
347
over the years save 50% energy savings with larger construction projects.
348
So naturally, that begs the question, Gretchen, how come people aren't installing these
349
everywhere? And I'm excited to hear what Neta says
350
about this one as well. But like, the quick, simple payback does not tell
351
the full story of a construction project, right? Like, if it did, then you and
352
I would be doing improvements to our living spaces all the time
353
because the payback would be the reason we would just do it. But a good
354
reason, I spoke with a grower in the Mid-Atlantic region who still uses
355
high-pressure sodium lights and time clock controls, and they
356
explained that they know every reason under the sun why they should install new equipment,
357
but the disruption to production is just a challenge to orchestrate that they—
358
I'd say the number one reason is operational impacts during construction. The
359
second one I'd say is these systems are not yet industry standard practice because
360
they're not required by codes or published in standards. But as I mentioned,
361
ASHRAE adding 90.1 language to require daylight responsive
362
controls means that, that because it's the basis for codes and standards around
363
the world, anyone could point to that and say, okay, Minnesota
364
chooses to adopt this. Okay, you know, Maryland chooses to adopt
365
this, or even the city of Chicago chooses to adopt this.
366
But it's happening in California at the state level right now. So I think that
367
if you're a California grower, This is the most important thing to know is that
368
it's not industry standard practice now, but by 2028, you're going to probably
369
be— January of 2029— required to do it. And then
370
lastly, upfront cost, right? Even if you get paid back, even
371
if you get a rebate, you've got to pay for materials, labor, commissioning,
372
training. And that's just a little bit too much for a lot of businesses to
373
tackle right now, especially during uncertain economic times. And,
374
you know, Henry from Agritecture posted, just today on LinkedIn about
375
how, you know, things are consolidating. So those
376
who are going to do these projects are going to be the, probably the larger,
377
more historically established companies. Yeah, that trough seems
378
to be a bit deeper than people originally thought. Yeah. Anything
379
to add on that? Gretchen covered it all. Like, I'm in complete agreement
380
that it is. First of all, you know, she mentioned California and oftentimes we do
381
see that states follow California. So that's something to say. I think it's
382
going to happen over time. More and more states are going to adapt what California
383
does and you see Historically, we've seen that in every industry, including the food industry
384
that I was in prior to this. So, I think that will happen. I do
385
think that interruption is really hard for them, and they just don't have the manpower.
386
Everyone's limited right now. You know, you don't have the manpower to make those changes,
387
and interruption to their day-to-day operation is a challenging
388
one for them. But we need to make this accessible too. I think that if
389
we can make it small and accessible and you start one room at a time,
390
it's possible to get there. It's just that you don't have to overhaul the
391
entire operation. You can start with one room and say, I changed
392
the lights in this room, or, I applied DLI in
393
this one room. What was the impact of that over the next 5
394
months? How did that compare to my last 5 months? And if that works, then
395
you slowly transition the change. Yeah, and I think it's important, and
396
maybe you can talk to this a little bit more, Neda, about this ability for
397
growers to understand that they don't have to rip and replace, as you say. And
398
I think they appreciate having the freedom to choose a mix and match
399
if that fits their needs in terms of sensors and tech. So how
400
does this help you think about how to approach established growers
401
who are set in their systems like the example Gretchen outlined?
402
Yeah, every greenhouse is different, right? Every greenhouse has different
403
needs. No one manufacturer makes the best sensors on the
404
market. We know that the technology
405
changes. So, I'd say it's, you know, our philosophy
406
has been you don't need to rip and replace, you can complement. So,
407
it's really coming in as a consultant and understanding from that
408
operator, what is it that you have today? Where are your
409
blind spots? Where do you want to go in the future? And
410
how can we complement what you currently have? So, in other words, let's say
411
back to DLI, let's say that You want to save
412
some energy on your lighting, but you don't have
413
DLI. You're worried about the expense and all the
414
wires that are gonna run around with all these different PAR sensors. How can
415
we address that? And the way we've addressed that is, what if you get a
416
PAR sensor that we can convert to a LoRaWAN wireless
417
so you can move your PAR sensor around? Now you don't have a bunch of
418
wires. Now you have flexibility to move the sensors around. Right. What
419
if you use wireless sensors for temperature, humidity? We actually have a
420
facility right now in Virginia that's doing some study on their blind spots. You can
421
actually deploy some wireless sensors to get to know those blind spots,
422
and you can still take that information, feed into your control system. Whether the
423
2 control systems, their existing control systems that we're not ripping and replacing, can
424
be integrated with microclimates is a different story. But if they can be
425
integrated, then the 2 systems can talk to one another. So really, it's
426
about giving them the freedom to choose what's right
427
for their operation. So I'm a strong believer that, you
428
know, growers shouldn't— they should really own their own
429
strategy, and they shouldn't have any limitations because of the vendor's
430
limitations. And I think it's really, really important that they have the freedom to
431
choose. That's helpful for that context. Thank you. So Gretchen, you did
432
touch on AI, and I'll give you both a chance to talk on it and
433
seeing what's coming up. But based on, you know, what you've seen in your experience
434
with this research, And obviously following the trends of what's happening on the AI
435
front, where do you see the biggest disruptions happening, or what should growers
436
be preparing for? Well, I want to reiterate what Neta said about a phased
437
approach. So no matter what happens next, whether there's this
438
great new innovation like a brain that will tell your control system
439
exactly what to do because it knows the weather for the past 50 years and
440
it's predicting the weather for the next 20 months, And that
441
would be awesome, but that's like also probably going to be like buying a very
442
expensive thing for a while. And so a phased approach, no matter what, will probably
443
be the best. And I think what, you know, the microclimate systems offer
444
is literal like card by card. So you can decide how much
445
equipment do you want to monitor and control with the system versus
446
letting your existing system continue to control some things and monitor some things.
447
And For example, with AI, if you want to first implement
448
a simple daylight responsive control algorithm and say,
449
turn the lights off when the PPFD gets too high, all right, just don't
450
overlight. Then you take a look at the crop, you say, that looks good. All
451
right, let's try now a DLI target, which is just based off of the target
452
you think I'm gonna reach at the end of the day. That doesn't require AI.
453
It could, but it doesn't need to. Folks from Cornell have been writing those algorithms
454
since the '90s. And those are just intelligent algorithms. But
455
AI, I think taking more of the agency
456
of making decisions might be where we start to see it happen, where it's like,
457
all right, the AI's gonna start to tweak your DLI targets day
458
by day because it's decided what you need. And we've allowed
459
that to happen with other things like screen operation. We've decided that, you
460
know, the control systems know what's best for the screens as they read
461
the climate. Responses, like what the temperature is and the humidity is, and they open
462
and close. So I do think that we'll start to see that integrated more and
463
more as we let the growers spend their time
464
doing the more human-valuable tasks. We've started to see the value of human
465
labor compared with AI labor, and that's where I think we'll start to see the
466
balance in the vertical farm and the greenhouses. What's important for AI to do because
467
it's cheaper, and what's important for the grower to do because it's much more expensive.
468
And Neda, what are you seeing from your side? First, I want to completely agree
469
with what Gretchen said, that phased approach, right? I love working
470
with Gretchen because we both have this mentality of things
471
don't happen overnight. It's a phased approach, and you want to make it
472
so it's accessible to the operator and it's scalable
473
so they can slowly work their way there. So from an AI perspective, I
474
think that AI is actually going to be very different than what
475
most people expected right now. A lot of people are thinking of AI as this
476
autonomous growing or yield prediction.
477
And yes, we're going to get there, but I think we're going to get there
478
slowly. And what Gretchen touched on is maybe making those
479
slight modifications in operation, right? To make those slight
480
modifications to your DLI or your HVAC systems,
481
AI needs to have the ability to Yeah. If
482
you don't have the data today and you're not collecting thousands of data
483
points, if you're not talking to an AI and actually
484
getting this— I think of the AI today as an assistant
485
that you hire that eventually is going to become your consultant or
486
it's going to do the work for you. It's either going to advise you, it's
487
going to do the work for you on your behalf, right? But if you don't
488
have the thousands of data points, if you can't summarize the trends, if you can't
489
make any recommendations and alerts, and you can't generate those reports based on the thousand
490
data points, how is this AI How am I going to actually get to know
491
you? So first of all, we have to step back and say, are operators actually
492
collecting data? We already said early on in this conversation they are,
493
but they don't have enough environmental visibility, which means they actually are not collecting enough
494
data, number one. Number 2, is the data coming together in
495
one place, or are these all different silos and the data is just
496
disparate and all over the place? Are you able to pull the data into one
497
place? You should be able to. So that's number 2, is that you don't want
498
your AI to be working inside of those. You want your AI to step back
499
and look at your entire operation and make decisions for you. Number
500
3, can your AI agent talk to other AI
501
agents? Some can, some can't. So you need to have an AI agent
502
that can speak to other AI agents. And then number 4, I'd say, which is
503
the long run, is can your AI with your environmental
504
automation now be able to talk to, let's say, your ERP systems?
505
And that's where we kind of move towards this yield prediction
506
and autonomous growing is when we really first have
507
thousands of data points. We understand it. The AI has
508
conversations with you. I mean, all of us are using chat now, right? A year
509
ago when we were using chat, we didn't trust it. And
510
now our chat or Claude, whoever you're using, knows
511
more about you, the way you look at things, the way your
512
operation operates. Yeah. And now it can advise you. But none of this
513
stuff can happen overnight. It happens very slowly. And you need
514
to be able to trust that AI to make decisions on your behalf. So maybe
515
at first, it's an AI that's going to come back to you and tell you,
516
hey, I suggest you make this change to your DLI. But you have to have
517
a set of eyes and say, I trust you, I don't trust you. Or, I'm
518
just going to do a simulation. I'm just going to do it in an R&D
519
room and see if this works. So when we talk AI, I think everyone gets
520
really excited. And we get really excited as a software company. Of course, we get
521
really excited. We step back and say there's a reality of
522
how it's going to progress, in my opinion, which is more of an environmental automation
523
AI that's going to help you versus the yield
524
predictions. And if you, again, you don't have your data house in order, you
525
don't have your data, you don't have enough data, or you don't have the data
526
coming together, how is the AI ever going to scan thousands of data
527
points to make decisions for you? Yeah, and that's my marketing brain is always on,
528
as you know, Neta. So it speaks to like this idea of like, Auditing
529
or having an audit saying, how data ready are you? How data rich are you?
530
Something along those lines. Because when you talk about these thousands of data points, and
531
I'm sure Gretchen can speak to this, like, you probably have some folks run
532
the gamut of just like, oh yeah, I totally get what you say, or like
533
eyes wide open, like, I'm barely measuring one thing here. The
534
thought of measuring thousands is just overwhelming. So just one quick
535
follow-up for Neda is just with the influx of like awareness
536
around AI, Claude connectors. You know, I myself am like deep in like Claude code.
537
And so every— if you follow X enough, there's always like Claude for this, Claude
538
for that. So are you thinking as a company about making connectors,
539
MCPs? You know, not to get too geeky here, but like stuff that can plug
540
into Claude easily for folks that are already dabbling? Because I imagine a lot
541
of these, you know, smaller shops are seeing how they can do more with less,
542
and then naturally they're leaning into AI. I myself like I've had it build spreadsheets
543
for me. I've had it connect to my CRM and populate it automatically.
544
It's my first go-to now. This thing that I used to do manually,
545
can I automate it? So I'm literally got Claude in a second window, a second
546
monitor, always seeing how I can put it to work. And I'm curious how
547
you think about that. Yeah, 100%. I mean, at the core,
548
you know this about us, Ari, is that we are an integration company. We're
549
an open company. core, the philosophy of Microclimates
550
has always been about not living in silos. So absolutely,
551
we are definitely looking at ways of— and we're already working on some
552
things on AI— is how do you get it to work with your existing AI,
553
your large language models, and then how much information can be
554
shared back and forth. And there's also the security aspect that always has to be
555
considered. So that's what our technology team is really focusing on, is the security aspect
556
and how do you keep that information especially now that you're sharing all of a
557
sudden environmental data, right? We've always believed that you
558
own your data. We don't own your data, which is why we're an edge company.
559
Your data is on-site, on-premise, not in cloud. You own your data.
560
So we gotta be really thinking hard also about the security aspect of it. And
561
that's what our technology team is focusing on. And it sounds like, Gretchen, when it
562
comes to security, that's something that's probably near and dear to a lot of growers.
563
Well, yeah, and it's near and dear to my heart. My mother is, Actually, like
564
one of the kind of mavens of IT security from the '80s. So,
565
she worked for a defense contractor in DC. She raised me to not give
566
out my data on the internet, not talk to strangers, you know, all that sort
567
of stuff where it was like, security is paramount, like operational
568
security for business. When we worked with one of the field demonstrations, Neta, I
569
think it came up that one of the IT teams was like, whoa, whoa, whoa,
570
you're gonna have a gateway and you're gonna be needing to connect to the internet?
571
Like, we've got issues with that. And so, I mean, even just that sort of
572
stuff, even 3 or 4 years ago, now it's funny for me to think that
573
we have companies that are almost divulging so much to
574
companies that aren't even their company. So I might recommend
575
continuing to be cautious with that phased approach, as well as considering making
576
enterprise AIs that are owned and managed by your enterprise, and
577
perhaps not sharing everything with external AIs if you
578
are able to avoid that. I know that there are potentially with like
579
microclimates, there's going to be ways for you to have AI recommendations that
580
are essentially always guaranteed to not be being used
581
to train other growers, for example. I think that was another concern that comes up,
582
and I want to make sure that continues to be something, you know, credible companies
583
do. So a grower I respect once said, I expanded one
584
acre at a time. And I think that person is still in business
585
and will probably be a good guide to think of as we all adopt more
586
automation and adopt more AI. I think everyone should explore their options and
587
find tech that allows for integration of existing systems, whether that's existing
588
hardware, existing enterprise software, existing AI
589
agents. It should be easy to install and it should have a low
590
subscription cost, which I think is something we haven't talked a lot about. But whether
591
it's AI or whether it's a suite of monitoring and controls
592
tools, those all have a cost, ongoing cost these days. It's
593
rare that you find something that you buy it and it's yours now. So I
594
think that the controls market is competitive though, so there's a lot of price points.
595
I'm excited to see that we continue to maybe use those levels of sophistication
596
to help people find the thing that's like right for them. But for me,
597
I personally don't use AI yet. I am maybe going to be swept
598
along in the wave eventually, but as an energy person, for me, I just
599
don't find it something I want to use yet. But I want to support whatever
600
systems people use, whether they're an industrial, ag, or commercial business. Because
601
it's not like, Annette, like you've said, it's not up to me to decide how
602
someone chooses to grow or how to choose to use their data. So as
603
we get close to wrapping up the conversation, Gretchen, I'm curious, when you
604
have conversations with companies in the space who are dipping their toe
605
in the sensor space or trying to revamp legacy systems
606
or hearing these conversations about AI and daylight control sensors, and, you know, a lot
607
of it can start to be overwhelming. So for folks looking to get a
608
start or get a foothold here, what do you usually recommend? Well, I
609
recommend finding free training. There's a ton of amazing free training
610
available from almost a decade or more now of free
611
webinars, short courses by Glaze, the Advanced CEA
612
team, the Indoor Ag Science Cafe. You don't have to
613
go in blind and get sold something by, you know, someone at a
614
trade show. You should get to know people, build relationships, and find out
615
like what's been proven in growers like you. So for example,
616
I've got a grower in California who's trying out root zone heating,
617
and they grow strawberries, and that's not terribly common yet with strawberries.
618
And so I think a key piece for, you know, persuading that grower
619
to try it out was getting utility rebate support,
620
showing them some academic studies, and making them feel like
621
they can continue to talk to other growers who do it. So those would
622
be some of the things I say about building trust. I think that the market,
623
ultimately, people need time. Sometimes my projects take years
624
to come to fruition because there's other things that are going on, like a pest
625
management thing or a delivery distribution thing and
626
trying to get new uptake agreements. So I see
627
things in like long years, much like construction. You just have to kind
628
of not see it as something that's going to turn around in the next 6
629
8 weeks or something. So, Nada, what do you think? 100%.
630
I think it's built over time. It's a progression. It doesn't happen
631
quickly. Yeah, continuous improvement, like kaizen. You know,
632
I think the growers that stand the test of time don't adopt
633
big rocket ships and then go to the moon. Most folks
634
are still on the ground, and someone called it a 7-day farmer.
635
I liked that phrase where it's like they are— that is what they're doing. And
636
that's what they're invested in. And it's like potentially this work, this
637
automation, this environmental visibility that is
638
amongst their priorities, but it's neither urgent nor the top importance.
639
So that's why we have to take our time to find when does it become
640
important. Oh, data centers caused your utility rates to go up by
641
25% and, you know, demand charges have gone up too. Now it's probably
642
the time to do an audit and improve lighting. So to that,
643
Neda, how do you think about conversations with new prospects and
644
people new to understanding this, new to understanding if this is even something that they
645
need? How do you usually start those conversations given everything we've talked about
646
today? New to understanding if they need environmental monitoring or controls in
647
general? Yeah. Oh yeah, I'd say if you're a new operator
648
and you're starting out and you have a greenhouse operation or you have a vertical
649
farm, whatever it may be, at bare, bare, bare minimum, we always say you need
650
to have some monitoring information. Right? You've got to have— you got to
651
understand what your crops are actually feeling and what they're experiencing.
652
And if you have one sensor in a greenhouse, it's just
653
not enough data point. So I'd say to a new operator, I'd
654
say at bare minimum, start monitoring. You don't necessarily need to jump from monitoring
655
all the way to automation, right? The automation is like the ideal place, and then
656
AI automation and algorithms are the next best place that you want to be
657
at. But You can do a lot of this work with just having monitoring and
658
setting timers. You know, we've seen operations that work fine,
659
and they run for a time period at that level,
660
but then they do need to move up to more of a control where you
661
have inputs and outputs. So, you have inputs coming from data, from your sensors
662
that are going to force the output so that your system becomes smarter and
663
smarter over time. So, I don't think that you necessarily need to go and purchase
664
I don't believe, I truly do not believe that you need to start an operation
665
and invest in a $200,000 climate control system. I don't believe
666
that. I think that you can start off with a few thousand dollars and just
667
start monitoring, and then set your controls and automation, scale your
668
operation, go one zone at a time. You don't need to just all of a
669
sudden spend $200,000 and every single zone is fully automated. You
670
can start slow because at the end of the day, you're going to run out
671
of money and you're going to be out of business. So, it's not a good
672
way of running a business. Well, like that other grower, he used the first
673
acre to pay for the next acre. So it's like if you do it all
674
at once, you've essentially taken all that capital
675
out of what could be sort of like a green revolving fund where I'm like,
676
great, I improved zone 1. Now zone 1 is costing me less money to operate.
677
I now have some savings to apply to zone 2 because becoming a
678
multinational grower wasn't done millions of acres at a
679
time. Yeah, that approach of going 1 acre at a time, Gretchen, is interesting.
680
How should growers think about that? Do they need a dedicated R&D
681
space for these sorts of tests? Can they do it with a sectioned-off area
682
of their existing growing space? I'm curious logistically how that would work out.
683
It's easier for indoor farmers to do things like that, I think, because they often
684
will have specific control zones that are already very well
685
separated, have different equipment serving it. Greenhouses often will have to put
686
up makeshift barriers if they want to set up a lighting zone of control that
687
Like Cornell creates these, you know, T's where
688
they have 4 different lighting treatments happening in one zone that might normally have
689
one treatment. And then you might see that in a commercial area as well, that
690
they start trying something out and have to build up makeshift walls. But with huge
691
greenhouses, that's not going to be possible. So I think
692
that's why a larger greenhouse would try it out on a smaller one first, and
693
then, for example, implement root zone heating or implement energy monitoring
694
at whole facility. But even if a grower doesn't have an acre, it's like one
695
zone at a time, I think, is where it comes to every size grower. One
696
room at a time. We do that at home, right? We don't renovate our whole
697
house at once. Generally, that would be extraordinarily disruptive and we'd have no money to
698
go on vacation or do anything else that's nice. So yeah, I think we
699
should take it as we all want to continuously improve so that
700
we are resilient and sustainable. And that doesn't just mean that we feel good about
701
the environment. It means that we're here to do business next year. So yeah,
702
I love what Aneta said about what the plants are feeling because— I love
703
that too. I was like, wow, that's a really good phrase. They're living organisms,
704
you know, they're living things. And they can't talk, right? So
705
these systems, they get insight that allows us to not
706
just save money but to actually have like real, like you
707
said, like almost organic impacts. They can't talk and you
708
can add sensors to get them to talk for you. So you can have
709
leaf temperature. That's how I think of a plant talking back to me, right? Is
710
if I can't— if you can't talk, I love that what you said, Gretchen. If
711
a plant can't talk, can you have a leaf temperature sensor
712
that is going to talk on behalf of the plant and let you know,
713
this is what I'm feeling? And then can you take that information and
714
feed into HVAC system because the humidity is too high in the room? And can
715
you reduce that by 3%? And then it will talk back and say,
716
I'm feeling better because I am in the threshold that I like to be in.
717
Yeah, they will get to the point where the AI is actually speaking for the
718
plants. I think about those experiments when I was in grade school. They would play
719
classical music for one set of plants and heavy metal for the others, and the
720
classical music plants would do better. So is anyone doing tests with music
721
in greenhouses? I can't speak for that, but I bet you a lot of the
722
researchers that I work with talk to their plants because they have shown that does
723
result in better outcomes. For sure. I will admit that I do.
724
I will admit that I have a banana tree in my yard right now, and
725
for the first time in Seattle, it's actually flowered 3 different flowers.
726
Thank you. And we're not supposed to grow bananas in Seattle, but I have been
727
talking to my banana tree. I'm sure it's helping.
728
So Gretchen, I'll go with you first and then Neta, just closing thoughts on this
729
conversation, where we are, or maybe some thoughts about the kind of the space as
730
a whole. We did mention Henry posting, you know, status of
731
what's happening in CEA. So I'm curious your 2 cents on what you see
732
from your perspective. Well, at the beginning of the year, I predicted a year of
733
retrofits, and I think that is what we see
734
happening and consolidation. And that doesn't necessarily mean,
735
you know, rocky situation for everyone. Ultimately, for Neta
736
and I, it means that we can work with people to, like, I think in
737
Henry's post, he said like a fairly well-built greenhouse is a very, very valuable
738
asset always, regardless of what happened to the company that owned it. So,
739
you know, times may change and the owners may shift, but we're there to help
740
that greenhouse become better. And so I think that is going to be an opportunity
741
we continue to do more as the vertical farms as well. How
742
can we get more monitoring and help that overhead costs go down so that whoever
743
takes that asset is going to have overall an asset that
744
is very profitable? And I also, for the rest of the
745
year, I see For my point, I'm going to be at GreenTech Philly. I'll be
746
doing a talk on the opportunities that are being presented by new
747
energy regulations and how that might allow for growers
748
to start exchanging energy with, you know, unique type of buildings like data
749
centers and others. So that could be a cool talk. And overall,
750
I would hope to share soon the results of the NYSERDA research that all the
751
glaze researchers worked on for the past 10 years. So that's going to be pretty
752
exciting. Those are kind of the 2 things for the rest of my year. Okay.
753
Thank you. Nada, what's on your radar? Yeah, it's, you know,
754
Gretchen mentioned the beginning of the year, her prediction. I'd say for the past
755
2 years, our prediction has been that we're going to— integration is going to be
756
the buzzword, and we're beginning to see more and more of that. Certainly, it
757
was a buzzword at Indoor@Con, and there was a panelist
758
discussion with our CTO that was involved that talked about what does it mean to
759
be integrated, Why is that so important? So, I think integration is going to
760
be the ongoing theme for a long time ahead. It's just
761
necessary. So, I think that's going to continue happening, and it's already happening for us.
762
We're going to go down that path even further. We're beginning to see more and
763
more companies that have the more legacy control systems changing
764
their models even and opening up their APIs and making it a lot more accessible
765
for companies to integrate with them. So, I think that thing is going to
766
continue. The other prediction that we've had for the past couple of years, and we're
767
just now starting to get there, is that these LoRaWAN wireless
768
sensors are going to take off in this industry. They've taken off in
769
other industries, but in this industry, it's sort of been, we've had some systems, you
770
know, you've got the Aranet systems, the Arroyo system, you've had other systems in the
771
market, but it's really gonna be about these open platforms
772
that the customer has freedom of choice. And we're hearing this over and over from
773
customers, and they get really excited when they take a look at our website and
774
we have 10 different sensors you can choose from, temperature humidity
775
sensors. You don't have to— you can pick and choose from 10 different vendors, 10
776
different manufacturers. So I think that theme is also going to continue,
777
that there's an excitement for these customers and operators to have
778
the ability to pick and choose what's best for their operation
779
and change vendors if they need to because a new
780
product has hit the market. So we've done a lot of integrations this year. It's
781
been really exciting, new integrations this year. Well, I appreciate that
782
feedback from you both because it seems like you both have a finger on the
783
pulse in your respective spaces about what's happening, where things are headed, because you're on
784
the ground and you're doing the work, working with growers. And I love to see
785
these types of partnerships, and I'm sure there's a lot more happening. So if there's
786
others that I'm not aware of and you need to bring them to my attention,
787
we'll get them on the show as well. But, you know, to see you guys
788
working together is really exciting because you're both bringing your respective
789
specialties and strengths, and it's so It feels like a 1 1 3
790
result here. So I appreciate the work both of you are doing for this space.
791
So Gretchen, best place for folks to connect with you if they want to learn
792
more? I'm on LinkedIn, Gretchen Schimmelfennig, and
793
other social media. I have other lives.
794
And Neta? Same here, on LinkedIn. And you can always go to microclimates.com
795
and schedule a meeting with me directly. Okay. We'll make sure all those links are
796
in the show notes. Thank you both again for an engaging conversation. Thank you. Thanks
797
so much, Ari.