Episode 32

Build For The Customers Of Our Customers - Fabio Rosa - TagoIO

Fabio Rosa, CEO and founder of TagoIO, talks about what it takes to build an IoT platform that scales globally while staying grounded in customer needs.

With over 1,000 supported devices and a GitHub-driven ecosystem for integrating LoRaWAN sensors, TagoIO has become a cornerstone in the IoT space. Fabio shares how his team prioritizes support for every user—whether it’s a student running a free account or a company deploying tens of thousands of devices.

He explains why TagoIO is designed not just for developers, but for the customers of their customers—making it easier to deliver value all the way down the chain. The conversation dives into the hidden costs of using AI in IoT, especially when querying massive datasets, and the steps TagoIO is taking to balance innovation with operational sustainability.

Fabio also reflects on key lessons from running the company: build fewer features, listen harder, and focus relentlessly on solving the right problems. He discusses how AI can be used not just to improve the developer experience, but to help end users extract actionable insights from their data—if it’s implemented thoughtfully.

Throughout the episode, Fabio emphasizes the importance of trust, transparency, and customer success obsession as guiding principles for long-term impact in a rapidly evolving tech landscape.

Fabio on LinkedIn

TagoIO Website

Transcript
Speaker:

Today's guest

2

:

on MeteoScientifc's

The Business of LoRaWAN

3

:

is Fabio Rosa, founder and CEO of TagoIO.

4

:

Fabio leads one of the most widely used

platforms in the LoRaWAN

5

:

ecosystem, a platform built

not just to visualize sensor data

6

:

but to support

full stack IoT solutions at scale.

7

:

With over 1000 supported sensor types

and a community powered GitHub

8

:

integration model.

9

:

Tago IO makes it easy for anyone to deploy

LoRaWAN based applications.

10

:

What really stands out is Fabio's

relentless focus on customer success

11

:

not just helping developers,

but enabling their customers to succeed.

12

:

That mindset, paired with thoughtful

engineering, has helped Tago Io grow into

13

:

a globally adopted platform that powers

everything from small student projects

14

:

to professional deployments

with hundreds of thousands of devices.

15

:

In this episode,

Fabio shares insights on AI's role in IoT,

16

:

how to balance innovation

with real customer pain points,

17

:

and why sometimes, the best advice

18

:

you can give is to recommend

a different platform entirely.

19

:

This podcast is sponsored

by MeteoScientific, dedicated

20

:

to spreading knowledge about Loreto.

21

:

We offer tutorials, run demonstration

projects, provide access to a global

22

:

forum, and produce this podcast

all in pursuit of LoRaWAN excellence.

23

:

If you'd like to learn more, visit

MeteoScientific.com.

24

:

Now let's dig into the

conversation with Fabio Rosa.

25

:

Fabio,

thanks so much for coming on the show.

26

:

Excited to have you here.

27

:

It's great to be here.

28

:

Now I'm pumped to have you on,

29

:

because Tago is such a big part

of helping people visualize

30

:

what's happening with LoRaWAN

and with really with IoT in general.

31

:

Maybe we start off with how many sensor

types does Tago support right now?

32

:

Yeah, we do have some decoders

33

:

that are ready to go, mainly for LoRaWAN.

34

:

It's not all about LoRaWAN,

but when you talk about LoRaWAN

35

:

it's a little bit

more than 500 different types of sensors.

36

:

However, it's very easy

to add a decoder for their own device.

37

:

So if they designed into a device

38

:

that have a need for a new decoder,

they can do this very quickly.

39

:

Right. I mean, 500, that's a fair amount.

40

:

That's a good amount.

I think you have over a thousand overall.

41

:

So there's lots of other devices

you can put there.

42

:

Not LoRaWAN specific right.

43

:

Oh yeah. Yeah yeah.

44

:

And now we use a GitHub.

45

:

So what we are doing is

46

:

we are working for the suppliers

so they can add it by themselves.

47

:

It took a little bit of time.

48

:

But now let's say there is a major factor

in Europe that has a lot of one device.

49

:

And maybe we don't know.

50

:

I think we believe that,

you know everyone, but we always need

51

:

more new people

and we can send a disruption.

52

:

And if they have, let's say,

20 different types of devices,

53

:

they can do by themself, upload on GitHub.

54

:

And then we approve the PR and you'll be

ready to go so everybody can use it.

55

:

Oh very cool.

56

:

So there's this aspect to it

that's saying, hey, let's not gatekeep

57

:

wherever possible.

58

:

Let's make sure everybody has access

to the stuff. Exactly.

59

:

And I know you guys have great support.

60

:

I know when I first got on there

and had some problems,

61

:

Freddie came out and said,

oh, try this, do this. You missed this.

62

:

And, you know, all of a sudden,

everything works.

63

:

So that was super cool to see

just how proactive Tago is in making sure

64

:

that the kind of customer

onboarding piece went well.

65

:

What do you think you guys do

that is special in this world?

66

:

There's a couple other folks out there

67

:

who are saying, like, hey,

you can visualize data with our platform.

68

:

What's so special about Tago?

69

:

There are few aspects

and frankly, one is the non-technical.

70

:

Basically

what you just mentioned about support.

71

:

Instead of doing a lot of,

let's say, advertising, we focus on

72

:

spending our money that we have on that,

the best support you can give.

73

:

So if you have a free account,

if you are student you are testing,

74

:

we are going to do the best.

75

:

We used to say internally

that we're going to support this person

76

:

for their problem,

as if they would be our top customer.

77

:

We really do that, right,

because we learn a lot as well.

78

:

And that's one differentiation.

79

:

I know that a lot of companies

you think that okay,

80

:

I need to spend a maximum

20 minutes per ticket.

81

:

And that now the big customer is fine.

82

:

So we say no, no, no support everyone.

83

:

And that's something that people recognize

and help with the business as well.

84

:

But from the technical side, it’s

really about the full stack.

85

:

We really focus on continuing

to improve our platform,

86

:

so we will not be recognized

only by the dashboard

87

:

digitalization as our normal act,

but also a very strong,

88

:

solid backend with features

that are really relevant

89

:

that will be much easier for people

to come to Tago

90

:

IO and do the configuration

instead of going to AWS.

91

:

or, you know, different cloud out there

92

:

So it's much more really relate to IoT.

93

:

And of course you'll have a

the the portal, the 2FA, SSO,

94

:

a lot of configurations

95

:

that is not for them, the developers,

but for their customers.

96

:

So I think that's really different because

we really are trying to anticipate,

97

:

and I think we did a very good job there

to help our customers,

98

:

to have a lot of information

and features for their customers.

99

:

I think that looks like, yeah,

a lot of people do that.

100

:

We really do that right away.

101

:

And we made a lot of mistakes

that we did here.

102

:

But I think we refine because we've got

inputs from the customer.

103

:

It's not like we're looking around.

104

:

And yeah, that doesn't have to

because other people are doing that.

105

:

No, it's really because

we talk with the customers and we try out.

106

:

Got it.

107

:

So one thing

I'm hearing there that comes through

108

:

in all of these conversations

on the Business of LoRaWAN podcast is

109

:

when I ask people like,

what's so important?

110

:

They say, oh, you find a business problem

and you solve it, and I can see

111

:

TagoIO is taking that a step further

and saying, hey,

112

:

we found this business problem

and we're going to solve it

113

:

not only for our customers,

but for their customers as well,

114

:

so that

115

:

everybody in the whole kind of line

gets some value out of what we're doing.

116

:

I can see that being super sticky

and really helpful to your customers.

117

:

There's thinking kind of beyond

just what they need,

118

:

but also what their downstream customers

need.

119

:

Yeah, that's a good point because we,

you know the customer obsession?

120

:

We use like customer success obsession.

121

:

So the little or small difference here

122

:

is that our concern

is really beyond the customer.

123

:

We want to make sure that they succeed.

124

:

The way they provide information

or service

125

:

to their customer

and is exactly not the same thing.

126

:

Because sometimes I say, yeah,

you know, this customer

127

:

maybe need to do things differently

or frankly, sometimes

128

:

should not be working with us,

should be working with somebody else

129

:

or doing something different. Right.

130

:

Because it'll be better for you.

131

:

And people say, really?

132

:

Are you saying that, are you okay

133

:

if I go and do other actions, say, yeah,

I believe that will be better for you.

134

:

We do this a few times

because that's our goal.

135

:

Will make sure that,

they can trust that we have.

136

:

And I'll try to help it out.

137

:

And it's always fun to be a good guy.

138

:

And say, hey, here's what I think

is going to be really good for you.

139

:

And that almost always, or in

my experience, has always come back to me

140

:

in some way, shape or form,

whether it's the satisfaction of going

141

:

and seeing

that someone just went out and crushed it,

142

:

or it comes back to you

143

:

and some other piece,

that's a super cool perspective to have.

144

:

You know, MeteoScientific and TagoIO

are in a very similar business

145

:

of helping people with IoT.

146

:

We do it in different ways,

and you guys are

147

:

of course

much more developed and much bigger.

148

:

But one thing

I think both of us are seeing

149

:

is this wave of AI coming and saying,

okay, here's what's going to be next.

150

:

And I'm sure both of us are thinking,

how do we incorporate it

151

:

into our businesses?

152

:

What do you see happening with TagoIO

and AI in the future,

153

:

whether that's six months,

a year or two years, whatever?

154

:

Yeah, I don't know.

155

:

How far can we go with our vision

because it's changed so often.

156

:

So I don't have this,

let's say the goal to say, okay,

157

:

I know what's going to happen or what

I really believe or that I don't know.

158

:

Sure.

But I can say what we have been doing.

159

:

We added, tools,

160

:

that will be running inside of TagoIO

basically two things.

161

:

One is to speed up the development.

162

:

Okay, really to help the developer.

163

:

For example, today

you have a search inside of our admin.

164

:

With AI you can do better search.

165

:

For example.

166

:

You can ask, even if you do a typo,

you can get a better result.

167

:

Or if you want to see a the tags

or location can do questions differently.

168

:

So also you can use the AI to write

169

:

because TagoIO is simple for one way,

but also powerful for the other.

170

:

And for the powerful part it's necessary

to write a little bit code here and there.

171

:

So the AI is helping there.

172

:

So yeah, we already have this version.

173

:

And by the way, we had the webinar

yesterday about how to use these tools

174

:

or the and the other part is use

AI to get insights

175

:

to make sure that our customers

176

:

can offer that solutions

to their customers.

177

:

Right.

178

:

So for example, do analysis of the data

that they have from their devices.

179

:

So that's not really during

the development phase,

180

:

but during the run time.

181

:

So we add these tools

182

:

and we also add to the website

recently a model context protocol.

183

:

That's up to 100%.

184

:

So people are very happy for that

but is very beginning very early stage.

185

:

So we didn't know what's going to happen.

186

:

But we see the adoption

and frankly we are using a lot too.

187

:

We have a small team of developers

inside of together that help

188

:

customers day by day,

and they are using these and doing great.

189

:

So I think that's the part that I see

AI, there are other things

190

:

that I guess may happen

because as you know, AI is so expensive.

191

:

So how you make sure

192

:

that you can offer these tools

and to be reasonable in terms of cost.

193

:

Right.

194

:

So you don't have billions of dollars

like other companies.

195

:

Like they.

196

:

not the IoT companies,

but in the AI domain.

197

:

Sure.

198

:

So we are offering these tools,

and collect the feedback.

199

:

Got it. Oh, interesting.

200

:

Interesting that you're saying

that AI is so expensive.

201

:

It has seemed to me, as a single user

that it's fairly cheap for what you get.

202

:

I mean, even Grok4 Heavy at 300

bucks a month, it's like, man,

203

:

that's a pretty competent developer.

204

:

I can't find a competent developer on,

you know, Upwork for 300 bucks a month.

205

:

Where are you guys seeing

that it's most expensive or where

206

:

are you guys having

to be most careful with it?

207

:

Oh, that's a great point.

208

:

You know I totally agree with you.

That's excellent.

209

:

So if you use AI in our experience

210

:

and as you said, pay 200 or $200

211

:

to help you during the development.

212

:

Great. Right.

213

:

Because maybe you can have just one

junior engineer or you

214

:

with a lot of experience in your business,

but then not coding every day.

215

:

Yep. Right.

216

:

So great acquisition, great ROI.

217

:

I was referring more the costs

when you start to do analytics, okay.

218

:

Because if you have let's say

1,000 sensors, that's not too much,

219

:

1,000 IoT sensors,

and they'll have data for years.

220

:

And collecting data every minute,

221

:

you're going to have billions

or trillions of data points.

222

:

So you may be tempted to do a prompt like,

223

:

Show me what happened last year

with this compressor, and the cost

224

:

from the AI perspective

and also from our servers

225

:

because you're gonna extract

billions of data points.

226

:

So it's a little bit tricky

how you do that.

227

:

And we want to make sure

that we could break this into parts

228

:

and even send a notification for you say,

hey, Nik,

229

:

do you really want me to analyze

everything here

230

:

is going to cost billions of data requests

and you just say “Oops,

231

:

I didn't know that.” So maybe you needed

to prepare the data in chunks.

232

:

So you having maybe the results

from every day while every week

233

:

then you have the aggregated data.

234

:

So that's the part that I have seen

AI helping as well.

235

:

But mainly humans right.

236

:

Today developers are like

okay okay I need to be careful here

237

:

because otherwise can be so expensive

to run.

238

:

It is simple things.

239

:

Yeah, it is

240

:

pretty wild that once you start using it,

you just want to apply it to everything.

241

:

But in many cases you can just write,

242

:

you know, some quick Python

that will do the job that you want to do

243

:

without doing anything with AI,

or you just use an LLM to help

244

:

you write the Python and that's it.

245

:

Then you send it, you know, fired off

and get the data you need.

246

:

Okay.

247

:

Super cool.

248

:

Is there anything else

that you've learned?

249

:

I mean, you've been running TagoIO

for a while.

250

:

One of the ways I think about this

251

:

is if you were to start over today

252

:

and someone said, okay, you've got to

go into IoT, what are you going to do?

253

:

We're going to start and you're

just starting with what you got.

254

:

Where would you start?

255

:

Where would you focus your efforts?

256

:

Yeah, that's there's not a fair question.

257

:

No. It's much easier now.

258

:

Right.

The scenario is very different. Yeah.

259

:

You think that now we are better

than a few years ago I think.

260

:

Of course we are still inside.

261

:

But also the competition,

the scenario are more complex.

262

:

I think one of the key things is, and

263

:

I think it’s very standard, right?

264

:

I think at that time

maybe we created too many things.

265

:

I think we really do like

266

:

as a platform to do

just the basic very well.

267

:

Right.

268

:

That's kind of like a good advice

for all guys that I know that had success.

269

:

Right.

270

:

So hey, don't create too many things,

271

:

don't create things

that customers don't want.

272

:

The problem is,

if you're doing this at the early stage

273

:

of the business or the market, you know,

so if you just create,

274

:

let's say, visualization,

like we start doing

275

:

that more visualization at the beginning,

like ten years ago,

276

:

people say, well, that's great.

277

:

But the analytics behind they say, okay,

you can do analysis using these software.

278

:

And this is I want to do here. Right.

279

:

So we start to add analytics and so on.

280

:

So I the end of the day we went

and built a very big system.

281

:

And we had to make sure that it should be

working very well together.

282

:

Right.

283

:

I so I think now we probably would as

we are doing that, we spend even more time

284

:

with our customers to try to understand

what are there real pain points.

285

:

Right. I think we did a good job there.

286

:

But I think what I would do

even more today, at the beginning,

287

:

the problem is that sometimes

when is not too much,

288

:

you may not be doing

and not doing nothing right.

289

:

So yeah, there are there are the pros

and cons, so I'm not quite right.

290

:

I don't think about that, to be fair,

291

:

because I don't know how much it is

going to help me right now.

292

:

But I think listen to customer

something that the were doing in that

293

:

I suggest if I start again, that probably

I would do a little bit more interesting.

294

:

And it's such good advice,

even though it seems simple to you and I,

295

:

and I think we have basically

the same problem, just very different

296

:

amounts of scale is

I'll do all kinds of projects all the time

297

:

because someone will say like,

oh, can you do this? Like, oh yeah, sure.

298

:

And I go off and do it

299

:

and you end up with 900 projects that now

300

:

you have to interconnect

those 900 projects

301

:

and they all have to work together,

and one project

302

:

has to be able to talk to the other.

303

:

And I think there are

there are many cases where, as you said,

304

:

if you just focused on making sure

the customer is getting what they want

305

:

and you're not building

just the thing that you think

306

:

is kind of fun to build,

there's got to be a nice balance there.

307

:

And it sounds like the piece of advice

there is really focus on the customer.

308

:

Stay as focused as you can.

309

:

Super cool there.

310

:

What's, what's next for Tago?

311

:

What do you see

kind of coming down the pipe in the next

312

:

like six months to a year?

313

:

Yeah.

314

:

The key things is really the AI. Right.

315

:

AI to integrate this in a way that would

be really very helpful to customers.

316

:

And the second part

is really to bring out these tools

317

:

so that it can help them

to have a competitive advantage,

318

:

because that is for me,

if you think about AI,

319

:

that's the key thing

that I always think every day.

320

:

Frankly, if I think about one, thing

AI is about how do we create

321

:

competitive advantage at the long term?

322

:

And, I don't know.

323

:

I don't think anybody know

the answer. Right.

324

:

Because it's like BlackBerry.

325

:

Well, 12 years ago and at the beginning,

you had the BlackBerry.

326

:

Great. Could go faster.

327

:

And few years later everybody had one,

like something like a iPhone now.

328

:

So what's the competitive advantage?

329

:

So we are really, again okay,

330

:

customers and,

we are in a stage that we are adding the

331

:

AI to help them

to create competitive advantage.

332

:

So customization is a big effect, right?

333

:

So that's the kind of thing

that is very hard to do.

334

:

I don't think it’s hard to add

AI to do something,

335

:

but to help them to create the environment

that they need.

336

:

If the customization, that's a big thing,

for example, it's all about

337

:

if you go to target, you say, hey,

I have this device in my hand.

338

:

And, I want you to do, cold

339

:

chain management,

create a dashboard for me with the alerts.

340

:

I tell you right now, it's

when they're doing in our version

341

:

that we are testing inside the target, or

we want to give it this to the customer.

342

:

So you do that right

in a way that they can use it and

343

:

they can work well and that they can debug

because the challenge.

344

:

Right. So you see. Yeah.

345

:

Then you need to debug this stuff

because it's not a toy.

346

:

Some people come in to target you to test,

347

:

to do something

very simple like in some makers studios.

348

:

That's great.

349

:

But maybe you will come to us to just

a professional applications that you're

350

:

going to be running very important things

for many, many years ahead.

351

:

So how you make sure

that is really valid, robust,

352

:

which I that's the thing

that we are going to spend a lot of time.

353

:

So that's what my vision is really

I had when I went up

354

:

and my team you put to the effort got it.

355

:

So one way I'm thinking of is

we kind of wrap this up.

356

:

But one of the takeaways I've got

is that I think of Tago.

357

:

IO is has really the execution

piece of a strategy.

358

:

And so a business comes to you and says,

hey, here's what we want.

359

:

You guys figure it out and return to us

the information that we need.

360

:

And there's so much in there to to say

like, okay, here's the device you want.

361

:

Here's how the device connects

to the internet. Here's

362

:

the backhaul piece, here's the decoder,

like all of these different things.

363

:

And at the end of the day,

364

:

the business that comes to you

doesn't have to deal with any of that.

365

:

It all generally just works as you

plug it in

366

:

and they get back

the thing that they want,

367

:

which is in this case

called chain management.

368

:

I mean, I'm talking to

I think Fedex tomorrow or the next day,

369

:

they're super interested

in this whole like, how do you make it

370

:

so that a normal customer who's shipping

cold desserts and is not technical,

371

:

how do they know the cheapest way

to do it, the fastest way

372

:

to do it, like the normal way to do it

without spending $400 a box,

373

:

which is what a drug company can spend,

but not a small dessert company.

374

:

So it seems like lots and lots of people

are specifically interested in cold chain,

375

:

but also just this

just give me a solution.

376

:

And that seems like

what Tago is really doing.

377

:

Well, yeah. Yeah.

378

:

We see about 85% of our customers.

379

:

They go to our portal,

they read the documentation.

380

:

I mean, they walk in account and read

the documentation, almost an example.

381

:

And they created their solution

not only for the device

382

:

but even

for hundreds of thousands of devices.

383

:

And the many times they not even talk to

us, they said, oh, we are good.

384

:

I'm here in Japan, in Singapore,

I have my applications running. Wow.

385

:

But about 15% of them really

386

:

want to have a relationship,

want our help.

387

:

Not like a consultant,

but someone that helps them

388

:

to create an application on top of TagoIO

389

:

and potentially

give them a referral, right?

390

:

They may ask, hey,

I need to hire an engineer like that.

391

:

You know, someone then say,

okay, take a look at this one or that one.

392

:

So you're right,

393

:

these type of problems and that many times

they are not expert in these domains.

394

:

And that's the beauty of the IoT

in my experience.

395

:

That's why I really like this.

396

:

IoT is a lot for the mechanical engineers.

397

:

It's a lot for the technicians,

for the managers.

398

:

Right. Do they have restaurants?

399

:

They have a machine.

400

:

And the data understand

about all this stuff

401

:

and a lot that they don't need to check

if they have the right partners.

402

:

So yeah, right.

403

:

They connect twice a few times

to have a deep conversation.

404

:

And then they try to get a round

of the IoT, you know, what are the best

405

:

you guys do for connectivity. Right.

Talk to me. Right.

406

:

Get this hired. There

he and there and so on.

407

:

So that's how I believe we're succeeding.

408

:

Yeah, yeah. And you guys are doing well.

You been around for a while.

409

:

You keep keep growing.

410

:

So it's super

good to see you out there. Yeah.

411

:

Thanks so much for making the time.

412

:

I know you're really busy

as a founder and CEO.

413

:

So thanks so much for coming on

and talking to us about

414

:

what's going on in the world of LoRaWAN.

Thanks, man. Thanks for having me.

415

:

That's it for this

416

:

episode of The Business of LoRaWAN,

I built this for you.

417

:

So whether you're a business owner,

a LoRaWAN professional or a hobbyist,

418

:

the intent is to give you great LoRaWAN

information.

419

:

Of course,

the best information doesn't come from me.

420

:

It comes from the conversations

421

:

we have with the people building

and deploying this tech in the real world.

422

:

And that's where you come in.

423

:

LoRaWAN is a global

patchwork of talent and ideas.

424

:

And ironically,

for a globally connected network,

425

:

most of the brilliant folks

working on it are connected yet.

426

:

Help me change that.

427

:

Introduce me

to someone awesome in your network,

428

:

someone doing meaningful work in LoRaWAN,

or just shoot me a name.

429

:

I'll take it from there

and get them on the show.

430

:

So we can share their work with the world.

431

:

You can always find me at metsci.show,

432

:

that's M-E-T-S-C-I dot

433

:

S-H-O-W, metsci.show

434

:

If you want to support the show

in other ways, you can subscribe,

435

:

leave a review,

share it with your corner of the world.

436

:

All those are super helpful.

437

:

If you'd like to support financially,

you can go to support.metsci.show

438

:

for both one time and recurring options.

439

:

We're also open to sponsors.

440

:

If your company serves

the LoRaWAN community

441

:

and you want to reach this dedicated

audience, let's talk.

442

:

I you want to try it.

443

:

LoRaWAN for yourself

create a MeteoScientific account

444

:

at console dot meteoscientific.com

and get your first 400 DC for free,

445

:

which is enough to run a device

sending hourly for about a year.

446

:

This show is supported

447

:

by a grant from the Helium Foundation

and produced by Gristle King, Inc..

448

:

I'm Nik Hawks.

I'll see you on the next show.

All Episodes Previous Episode

About the Podcast

Show artwork for The Business of LoRaWAN
The Business of LoRaWAN
Learn From the Pros

Listen for free

About your host

Profile picture for Nik Hawks

Nik Hawks

Incurably curious, to stormy nights and the wine-dark sea!