Orenda Coffee — More than just a failed Kickstarter
26th January 2019
This is work in progress. Although I have extensive records of events, some is from memory and may be incomplete or otherwise flawed. Because of the nature of parallel stories during this saga, not everything is in chronological order. I’ve tried to attach specific dates where possible.
Corrections to firstname.lastname@example.org.
Update October 2019
For a final update on this saga, skip to the end. I’d like to thank the thousands of people who enjoyed this story.
We all get pitched products that are too good to be true. Advertising on the internet thrives on it. More so on social media, which as we’ve learned, knows an awful lot about your preferences.
Sometime in 2015, I was shown a Facebook ad for a coffee maker. One that was IoT (Internet of things), fully programmable, had an app for different recipes, and a built in grinder. Even then, this wasn’t my first Kickstarter, or even my most expensive — although at $424 all up, it wasn’t cheap. I knew the risks, although at the time, each of these features talked to me.
Less appealing to me, but still selling points, the maker would perform coffee flights, and be easy to clean and do my laundry (well, no).
Not only was my standard Cuisinart maker on its last legs (more on that later), but the idea of programmable, freshly ground and roasted coffee was pretty appealing. I’ve owned a series of coffee makers, including two Keurigs, but I would never have styled myself as a coffee snob. What I am though, is an expert on embedded software, control and IoT. In fact, I worked for an IoT company at the time and wrote software that wasn’t very different to what would likely be controlling this.
The product was of course, the Auroma Coffee Maker, by the Auroma Brewing Comany (ABC), who later renamed themselves as Orenda, with the Orenda coffee maker. I’ll use “Orenda” here for both the maker and the company, although there are plenty of references to “Auroma” around the web. The project was run by a group who styled themselves the “founders”. Pretentious? A little.
Somehow, despite the risks, a fabulous machine which would do everything, $424 didn’t seem that much. In February of 2016, my Discover card was charged. I’ll talk more about this fateful choice later.
The project went as many Kickstarter projects do - problems with manufacturing, and the quality of materials. Delays due to suppliers and predicted delays. We were told the team had moved to China to oversee production and were basically sleeping on the factory floor to make it happen.
The red flags started in early 2017. There were months and months between updates. However, one of the first of the rogue’s gallery appeared, “Jade” who would respond, sometimes, to queries on Facebook, with vague promises that it was really happening. At this point, some individuals started asking for refunds. I have no evidence that any refunds during this period were ever issued.
As 2017 rolled on, one of the updates included a claim that the quality of the glass was poor, and had to be remade with consequent delays. This claim about the glass would come back to haunt them.
Finally, in a December 23 update, the news we had been waiting for:
Hope things are going well on your side. We’re delighted to inform you that your Orendas have been made, and are on their way to you.
This turned out to be a fiction and the first of the really obvious dishonest claims by the company.
It wasn’t until nearly two months later, on February 16th, we learned the truth. In a post replete with pictures as “proof” that it was really shipping, we were told that in fact there were QA issues at the last minute, and that they had to be repacked. The red flags now started to go up, and people started to question if this was a scam.
Despite the censorship of the shipping label, it was determined which ship the machines were on, and many of us enthusiastically followed its progress across the Pacific. The ship passed within miles of one customer in Hong Kong, and the port it stopped in, Oakland port is just a few miles from here, although the machines were then shipped to an apparent shipping distributor in Salt Lake City. The promised dates in this post, at least for the Kickstarter supporters, proved to be largely accurate, with customers in the United States receiving their units in March/April 2018.
The first customers started getting their machines. No, there was no app to begin with. Yes, it did make coffee, and took an awful long time about it through a complex recirculation mechanism. Yes, it’s a bit lukewarm, but the coffee is OK.
I shortly after received mine, and others as well started to see problems: 1/3 cups of lukewarm coffee full of grounds. Failures to dispense. Cracked glass. Missing parts. Obviously leakages from the top and bottom sections. The list was long. All of these items had obvious causes and solutions (the dispense failure for example happened after moving the machine and was a software problem with the strain gauge).
Below is a picture I took from later on, but this is very typical of the reports — noticed the messy grounds, 1/4 cup of lukewarm coffee, and general mess. And yes, that’s my Cuisinart on the right.
So, despite the founder’s overwhelming claims about care of materials, quality assurance, lots of testing and indeed, that glass, it seemed like they had a lot of serious problems in many machines.
But there was more — apart from the blatantly missing app — there was an extensive list of features listed on the Kickstarter page or shown in the pictures which were not there in customer machines.
In particular, missing were:
- An On/Off switch — A rocker design. Not strictly needed, but appealing to some.
- Flight capability — The ability to brew several coffee flights at once with different recipes — something boasted about in the patents.
- Swing-out arm — Something that might have made breakage less likely
- An LCD display — Which might have given a little more info than the LED feedback.
Yes, all of these could be quite easily justified in their removal. The problem is there was little or no heads up that these features would be missing. Worse, those original pictures would continue to be used on social media as promotion photos despite them obviously being wrong.
Kickstarter versus Website
As an aside, I will note that many later customers had no idea the product had originally started as a Kickstarter project. They had instead brought the product directly from the website, directed there by searches or dubious reviews. It appear that it was mostly the website customers who were the bulk of customers to not receive machines.
“The App” took on a kind of mythical aspect, since it promised so much. For example if we look back at a Feburary 2016 Kickstarter update:
We have gone through two additional iterations with people we know before we ask for your precious time. We are pleased to announce that we have worked through major confusion points, and finally have a workflow that works well in all use cases. Currently, our designer is integrating through the UI for the right feel, while our app developer is implementing the new workflow into the structure of the app.
Awesome, they’ve put a lot of thought into this, right?
When the iOS did appear in late March, it was not to be. It was fraught with obvious bugs, clearly bad UI flow, and ultimately not the magic fix that Orenda, whilst not strictly saying so, had suggested might a cure all for everything.
But here’s the problem: The app doesn’t really control the machine — yes, it feeds it parameters and tells it when to start and such, and reporting of the status is via the app, but the software that controls the machine is quite a different thing — it’s the firmware on the machine itself. And the firmware quite obviously had problems which the app was never going to be able to paper over. More on the firmware shortly.
Don’t get me wrong — the app is important for presentation, and setup of the machine and is ultimately the front end to everything — so it has to be done well, but there they failed too.
Ultimately, and as expected (at least, to me), the app failed to make the water hotter or avoid problems with not completing the brew cycle.
But where was the Android app? From an update from March 19th, 2018:
The Android app will be released on March 21st.
We have ONE bug left on Android that is still annoying us, but it’s better than letting it annoy you. We’re doing everything we can to solve it before sharing the app with you. This should not take more than 24–48 hours.
The Android app was not released until May 29th. It’s not really clear how 48 hours turned into 6 weeks, but again, this was now the norm. Many people pointed out that both the Apple and Android app store acceptance times were quite quick — sometimes only hours.
The Android app proved to be, if anything, worse. It would crash in certain setup cases unless stepped very carefully though WiFi setup.
Now, to be fair, the design of WiFi on IoT devices isn’t ideal — the devices first have to set themselves up as an access point, and then have another device connect to them to configure them, and finally, in a seat-of-the-pants moment, reconfigure themselves to connect to your local WiFi access point and then have you return the configuring device to your local WiFi.
This can mostly be smoothed over with good app design and flow, but again it was not to be.
After pressing for weeks for an RMA, Orenda customer support finally “agreed”, with the following:
May 30 2018 from “Michelle”:
An RMA has been created. We will keep you posted with the refund once
Fantastic! But wait, what? What is the RMA number, and where do I return it to? After much pushing and 5 weeks later on June 25:
The refund is not yet processed because the machine is not yet
returned to us. We already created an RMA last May 30th and we are
still waiting for the item to be delivered back to us. Once we have it
then we can proceed with the full refund.
Wow, OK. So they think an RMA was issued, which was the excuse for apparently ignoring emails on the matter for 4 weeks. Finally later that day:
Your RMA # is 111000022. Can you please provide us with your best
contact number so we can reach you.
Progress? Maybe. What do I do with that RMA number, and where do I send the machine? It took another 2 weeks of pushing. On July 4th from Isabel:
You don’t need to provide our address, just give this RMA #111000033
to FedEx for reference so that the machine will be delivered back to
us. Once we have the item, the full refund will be processed.
My partner, who literally ships a thousand items a year, took this to FedEx.
So FedEx had no clue about this number. I had a similar exchange with Eden later in the year, and explained that this was fake, but again, nothing of any consequence came of it.
Now, what’s going on? Could I have misread the number? Did I miss something from the emails? Well, I did very carefully check all that. The RMA number I was provided with appears to be an internal tracking number that has nothing to do with FedEx, and as you can see above, isn’t accepted.
But here’s the thing — someone else had a nearly identical experience, so it wasn’t just me being a FedEx return newbie.
Why would they do this? Someone either make something up (i.e, lied), or got the number wrong. The problem is, this is not normally the return process, but instead involves the seller generating a return label. If you’ve an Amazon return (which is UPS, not FedEx), you would have done something similar.
So what exactly is going on here? I really have no idea.
On Tuesday 22nd January, I got a reply from Pawin on this matter. He supplied me with screenshots from their system with my RMA request in it (in fact, there were two), along with screenshots from allegedly successful RMAs (something I’ve yet to corroborate).
The problem? Again, there appears to be nothing like a suitable FedEx number in what he provided me. He also suggested the number had expired, and that I ought to have tried it immediately — sure, except there was nearly 6 weeks between the RMA generating and someone bothering to tell me the internal number — which doesn’t appear to be relevant anyway.
In April 2018, I realized after some research, that the computer on board the machine was one that I could access and program. The computer in question is a “Particle” board, not entirely dissimilar to the venerated Arduino. It has a processor with a fixed set of I/O and C programming language setup with a library for accessing the sensors, actuators and other features. The software that runs on such computers is usually called “firmware” and happens to be right up my alley, although my background is much more complex computers typically running Linux.
The source of the firmware written by Orenda (or as it later turned out, a third party developer) wasn’t available for examination, so I hoped Orenda might consider open-sourcing it. As it happened, a phone number had been obtained by a customer, and I got a response to some text queries when I asked to do this. I believe this was Pawin, but I’m not certain. I ended up making a Google doc with a detailed response their reply that I could share with other customers. Here it is, with of course my reply in bold italic.
The firmware is not bad.
I write firmware, have done so for 20 years. I’ve looked at millions of lines of it, spent some years chasing down bugs in firmware in other people’s devices, and working around them. Firmware is very exacting, and can’t be “good enough”.
For the Orenda, I see multiple failure modes reported by users, including the failure of mine — all apparently firmware, rather than hardware issues. You told someone else there are security issues (presumably in the firmware). IoT security happens to be an expertise of mine, although I can only speculate at the nature of your problem.
So when you say the firmware is “not bad”, that tells me it actually needs substantial work. Prove me wrong if you like, but that is the overwhelming evidence here.
The app had issues, that’s for sure, most have been resolved.
Perhaps they have. But the app in this case is largely UI and interaction. No, I understand that apps for control are non-trivial things, I get that. But the firmware is the key piece here. Without that, the app will not, and cannot function.
What are the advantages of open sourcing it? Why do you believe that’s a good idea?
Your company is desperately in need of good will. I won’t repeat the reasons here, they should be painfully obvious. Open Sourcing it will give you a small bit of good will. It’ll encourage people to really get involved and take an emotional stake in it. And it’ll show you are doing something practical. Plus, with my blessing (or anyone else, I’m not hung up on ego here), you might claw back some credibility.
How can we open source the firmware without exposing and practically open sourcing the hardware too?
You published much of the internals of your machine in your patent. The electronics, I believe it’s largely based on design published and open by Particle. The rest? Someone can open the machine and look.
I don’t pretend to know your business model, but I think you might be protecting the wrong thing.
Won’t you need to understand everything about the hardware internals to do anything useful with the firmware?
A common misunderstanding. I won’t go into the details unless you want, but the answer is a resounding no.
Also if someone plays with the firmware in the wrong way, he could literally cause a fire by misusing the heater or other safety concerns.. have you thought this through in detail?
Excuse me if I find this a little fatuous and nanny state. I can cause a fire with any number of appliances in my kitchen — and most of those don’t have software at all. Or I can just go buy some matches. Give us a little credit. This is what warnings are for.
Let me add this. Right now, you owe me several hundred dollars for non-functional machine which you’ve been far less than forthcoming about. Adding customer support is very unlikely to address this. Opening the software might just get you past what looks like hundreds of pissed off customers.
Of course, there was no response to any of this, so I had to start from scratch.
To begin with, I was able to access the machine firmware via the Particle web interface, below.
While this is a little interesting, it doesn’t really get me anywhere. Without access to the Orenda source, I didn’t know what the parameters to the interfaces such as “brew” were, so there wasn’t a good starting point here.
As it turned out, it wasn’t entirely true that I needed to start from scratch— the base firmware provided by Particle themselves contains a great deal of foundation code and an interface to experiment with the peripherals. After a couple of weeks of frustrated experimentation (for the curious, I reflashed with the “Tinker” firmware), I figured out and documented about 90% of the internal functionality of the system. There are a lot of stock components as you might expect, with standard ways of talking to them from the Particle board.
For a coffer maker, it’s definitely complex. There’s 3 pumps, 2 motors (for shakers) and the grinder itself, a weight sensor, TDS sensor, 2 thermometers and a couple of valves. Also, for a mostly vertical machine, a surprising amount of horizontal tubing. More on that tubing later.
In order to further my understanding of the machine and its working, I did what any engineer would do, and took it apart. The number of tools needed was a little surprising — 3 different sizes of torx and 3 other screwdrivers. A lot of the construction seems a little flimsy and of the nature of ad-hoc work arounds to problems. Typical of a version 1 of hardware, but definitely ready for a version 2 rework.
You can read my full internals documentation in this Google doc. I was later able to complete the documentation to about 98% functionality as per the table at the end.
I admit it, although I’ve had a great deal of hardware exposure, I’m just not a hardware engineer. I took a lot of wrong turns figuring this stuff out, but I got there. The LEDs in particular, something you’d think was very simple, turned out to be complex — fortunately Particle provides a complete library to talk to them. Still a few items escape me though — how does the heating in recirculation work — note the “keepwarm” item in the early screenshot. Without taking the machine entirely apart or getting more details from Orenda, we may never know.
With this in hand, I was able to get to the interesting parts and start developing code to control the pieces. This proved to be very exacting, and taught me awful lot of the ways making coffee can go wrong — and ways the Orenda firmware as shipped was going wrong.
And no, in case you are wondering, I completely failed to set anything on fire, and only made a few small messes with water and coffee grounds.
One particular piece of hardware is particularly complex to use, and Orenda describe it here:
Generic load cell design for a scale is not very complicated. However, once incorporated into a fluid loop design, the embedded load cell can be a huge challenge. The reliability of the scale’s readings was one of our biggest concerns. Originally, we had encountered some problems with drifting in the load cell readings after extensive usage. The drift causes the weight to become unreliable in long periods of use. To counter this problem, we isolated each variable, and were able to point the causes of the drift. As a result, we are happy to inform you that the load cell no longer drifts!
Indeed, it’s a royal PITA to use. The load cell is used to weigh both the fluids and the ground coffee. It seems easy enough, but there’s indeed drift, as well as spikes as weight settles, and also needs to be re-calibrated at the start of each stage of brewing. On top of all that, the measurements vary wildly by temperature.
It’s from this that I determined that many of the machine failures were in fact exactly mishandling in software of the load cell — some threshold was not being met and in their firmware it often went wrong, and it would get stuck at brewing and then not dispense. One person reported that this happened after moving the machine to a different spot in the kitchen, again pointing to this particular mishandling.
I struggled somewhat with this problem too in my firmware. In the end I determined that the figurative belts and braces were needed for every step — not only weight measurement, but also strict timeouts — if for no other reason, to avoid flooding anything, which was entirely possible.
There was more though, and the 3 pumps onboard all had different flow rates, all also varied in speed with temperature of the waster — which is quite a lot, from about 20 degrees Celsius room temperature up to 100 boiling water. Not quite the extremes of space, but typical of a kitchen appliance.
Also in some concessions to practicality (I didn’t have a suitable thermometer and don’t own a TDS meter), I made some guesses about calibration of the thermometer readings, and in what felt a bit too much like homeopathy, highly diluted some salt water several times to get some reasonable values for TDS.
According to the chart on the Kickstarter campaign page these values are critical. Somehow my cheap coffee maker manages, however. But certainly the values could be much more precise, and a lot more logic put around variations in temperature, etc.
Making an App
As I noted earlier, not only was the interface between the Orenda firmware and their app a secret, their app was something of a disaster. Although I would have tried to write one if I had to, I reached out to the community for a developer. The person I found was Prasen Palvankar. He ended up doing a bang up job on understanding what I’d done and making a polished app.
Below is a screenshot of an early version. Since my machine at time of writing is not connected, I can’t show you much more, but Prasen thoroughly thought the the problems that I highlighted in apps with WiFi connection and other issues resulting in a much smoother experience than Orenda acheived.
Suffice to say, Prasen was very professional, and one of the most mature developers I’ve worked with.
Putting it all Together
With all the knowledge of the internals in hand, I was able put together some processes. The first thing I tackled was a complete flush of the water in the machine. This seems like an obvious feature, right? It’s critical for cleaning or if you want to refresh the water or even just move it. But the Orenda app/firmware have no such feature.
There are two cleaning cycles in the original firmware, but they are mainly to cycle water through the TDS loop and flush out any grounds. Using those to remove all the water from the system would take something like 20 cycles — not a great user experience.
Now, it’s not entirely trivial to do — as mentioned, the pumps are all different rates, and there’s a chance of overflow, so you can’t just turn them all on for a while and hope. Also, the sensor in the reservoir in the base is such that the first pump in the system needs to run for a while to determine if and how much water is remaining to be emptied.
If you watch the video carefully, you’ll see the water level in the brew chamber cycles to avoid overflow.
The next piece was to implement brewing. The Orenda firmware doesn’t allow brewing with your own grounds, or indeed, just making hot water — instead it insists upon the entire process and grinding beans — again, a seemingly obvious feature oversight. So here it is:
Wonderful, yes? No. The water tasted like rubber tube! What had gone wrong? Perhaps I’d overheated some tubing or plastic elements in my experiments — certainly possible. But surely this wouldn’t be possible with food grade components. Could it be possible that no one else had noticed, since the coffee masked the flavor of the tainted water? Again, possibly, but seemed unlikely.
Or finally, could be that Orenda simply wasn’t all that careful in selection of materials. That’s a serious accusation, but it would fit their pattern. I never got around to trying it on another machine, so we’ll never know for sure.
At this point, with my enthusiasm for the whole project waning, and my sick leave at an end, I gave up on finishing. Obviously programming a complete grind/brew cycle from this point would not have been difficult — operating the grinder and shakers is trivial enough, and also replicating the clean cycles too would be straight forward.
But apart from the tainted water, and also lack of complete knowledge of the “keep warm” aspect of the TDS loop, there was one more confounding factor here.
The Particle board authenticates itself with the Particle cloud. There’s two ways this can be done — with individual accounts or using a 3rd party backend via the Particle cloud. It turned out, after a great deal of back and forth with Particle that I had taken advantage of a hole in their system to even be able to talk to my machine at all.
As it was, anyone using the Orenda app to setup their coffee machine would then have their machine authenticated with Orenda’s Particle account and Orenda’s authentication mechanism. This latter piece of software was another item that Orenda never admitted existed, but clearly must exist given the Particle cloud architecture.
Put another way, if a customer had used the Orenda app first, they would be unable to use my firmware later on without intervention from Particle in removing their machine from Orenda’s account.
I was unable to get a clear answer from Particle on either the exact nature of the hole or possible mechanisms for general usage, so this put something of a damper on trying to make this a polished end user setup.
The source to the firmware is on github.
In order to capture some kind of picture of customer satisfaction, in May of 2018 I created a survey on SurveyMonkey. Although in the end nearly 150 people replied, it’s limited to 100 responses without fees, and the most critical of those are shown below.
With the understanding of self-selection bias, and other fallacies of surveys, it’s still pretty damning. That is, 5 people out of 100 had a satisfactory experience. Even accounting for hardware failures, frustrated customers and other issues from first cut hardware, from start ups, that’s an extraordinary high customer failure rate. It’s at this point that they certainly should have pulled the emergency brakes and asked some serious questions about what they were doing.
From the two-thirds not shipped rate, and some back-of-the-envelope calculations from the number of Kickstarter orders versus website orders, I determined that they were sitting on approximately half a million dollars in unshipped orders at this time. That’s not small potatoes. To this day, many customers still do not have anything, although I don’t know what that proportion might be without running another survey.
Around June of 2018, it was clear to me that there was going to be no quick replacement or fix of the machine, and Cuisinart with the rusted bottom had to go. I looked at various machines and based upon a USA Today list I looked at both the Moccamaster — a very sleek stainless steel machine, which by all accounts in reviews produced excellent coffee, which is however, north of $300 — and also a Black and Decker basic machine, which all up was less than $20 from Target that I eventually settled for. Yes, it’s a bit flimsy and can make a mess if you don’t seat the lid properly, but it makes coffee extremely quickly. And did I mention it was $20?
For these types of projects, the software (firmware and app) is a heavy investment, but at the same time, it’s often not really worth much by itself. The Orenda firmware (both mine and theirs) only runs on the Orenda, and has very limited value elsewhere. One could even argue that mine is more valuable since it’s open to anyone to reuse either on the Orenda or to take as ideas for elsewhere. And as a learning experience with this type of platform, it ultimately has personal value in my field.
Similar comments can be made about the app — although in the case of the one done by Prasen, the design is a lot more generic, so has a lot of value in a different contexts in terms of time saving. But in all cases, the software has no real monetary value, and there appears to be no software patents or other IP tied up in the software, which is one of the reasons I pushed Orenda to consider open sourcing.
Also, there’s no “razor blade” business model here, like with the Keurig pods, nor any subscription service. So the business here must be based upon profits from hardware sales and an eventual exit with the intellectual property, including patents, processes, designs and the like.
The patents in question are somewhat interesting:
- System and method for controlling the brew process of a coffee maker
- System and method for automated coffee grind refinement
But how much monetary value another company might eventual pay for either the ownership of the patents or as royalties in use of the ideas is an open question. There are certain many many ways to make coffee without these.
To put this all into perspective, investment companies like SOSV which invested in Orenda, support a whole slew of companies at once. The expectation is that most of these will soon fail, but a few will do really well. The investment company then invests more money into the few companies that did OK, with the eventual aim of an “exit”, either by having the company bought by a much larger one — an acquisition — or in rarer cases, an IPO, or listing on the stock market. For all the billion dollar “unicorn” tech companies out there that you know about, there’s uncountable many more that you haven’t heard of that fall by the wayside.
This is how investment companies make their money — lots of smaller bets and a few big payoffs. This end goal of both the investment company and the company founders can explain some of the extreme arrogance towards the customers in cases like this — the end goal is not to make the customers happy, it’s to make the company last long enough so the company can be sold to someone else for a fat profit. In fairness, very few companies treat customers with the arrogance that Orenda has done, but this is nevertheless the goal taken to an extreme.
It was repeatedly asserted by a number of customers that this was a Ponzi scheme. In short, it’s not — that’s an investment setup which promises returns to early investors by rolling the contributions of later investors. The SEC in the United States looks very dimly on such things, but this is not that — is arguably customer fraud, and different agencies are (or should be) involved.
So what was to be done to reclaim my money? Many customers turned to their credit card companies, with mixed success. A document was assembled to aid in this process.
One of the problems was the length of time (a couple of years) between the credit card charge and the claim of non-delivery, and because of this, many credit card companies simply refused to deal with it. Some customers had better luck with more high-profile companies like American Express.
Customers like me, however, who had received a machine were actually in a worse position. As I noted at the very beginning, I had paid with Discover, a company I’ve been with for a very long time. Discover’s response was, in my opinion, very shameful.
Not only was their response to my claim full of inaccuracies, such as repeatedly referring to the company as “Kickstarter Auroma” despite my many corrections, they refused to consider the fact that the machine was fundamentally different from that described saying words to the effect that “products sometimes differ in detail”, even after I pointed out that as a coffee maker, the product fundamentally failed in even brewing a single cup. They suggested I seek other legal recourse.
I later filed a much more extensive claim with them, pointing out the errors in their original responses, the extensive dishonesty of Orenda, and using a subset of the material for this article. Their response seemed arbitrary at best, and basically came down to (a) They didn’t feel like it and (b) Orenda would have tell them to issue a refund.
I might have understood if Discover had based their refusal on credible published policies or some kind of reasoning, but this was infuriating to say the least.
Like many others, I filed a complain with the Better Business Bureau, despite widely being panned as ineffectual, at least I can say I had done it and then move onto to next steps. I also filed a complaint with the California Attorney General and Federal Trade Commission. The latter I heard nothing from; the former, I did get an actual letter from, as several others had from their state Attorney Generals. Here’s what I received:
So, they could not really do anything without a real mailing address. As others had noted before I sent this, Orenda were using a virtual office with a location in Delaware — an address that did not accept any mail or packages, as one customer noted when he tried to return a machine there.
I did eventually start to prepare a second set of material to follow this up, but never got around to sending it, instead choosing to focus on this story. I may eventually do that now that I have all the material together. In particular, several phone numbers were gathered from customers who were called by Orenda, which may be enough of a lead for investigation.
I am many things, but one thing I definitely am not is a lawyer. I’ve successfully won a non-trivial amount in small claims court, and once spent a week on jury duty, but that’s it. So what follows is lay opinion only.
Many people suggested a class action suit. Here’s the problem I see with that — first, it would require someone to step up to lead such a case, or involve a lawyer who’d actually be interested. No one ever came forward as a likely candidate for either. And my expectation is that the money returned would be quite small — much of the money would go to legal fees given the relatively small number of customers compared to class action suits that I have seen. There’s no claim of harm for example, just loss of money, and many small claims end with credit against products from the company, such as a recent one against Vitamix.
A more credible path might be the small claims court. This costs about $75 to file, and some time in court. There are definitely cases where disgruntled customers did such a thing and succeeded. But you have to serve it to the right party — and they’re in China. Similarly, even if you won a case somehow, then how would the payout be enforced.
Now, it might be possible to put the claim against SOSV (who are in San Francisco), who are investors and strictly speaking own the company. However, my expectation is that the terms of investment are very carefully made to avoid just such responsibilities. But of course, I could be wrong here.
Facebook Fake Likes
Not entirely unusual, but it seems like Orenda hired a marketing firm to pump their social media presence. You don’t just get 10 thousand likes by accident. The fact that both values are just over 10 thousand is suspicious all by itself. I’ve not looked, but I’m sure there are companies that exist to do this precisely for you.
And of course, having family and friends comment and like on your projects isn’t entirely unusual, but the following is a bit strained.
Sorana is of course, the sister of founder Orinicha. You can find similar likes on the Kickstarter updates.
Censorship and Forums
Originally, Orenda had started its own forum to allegedly help with customer issues. Although there were hundreds of posts to the forum, there was not a single instance where they provided any direct help to any customer, instead directing them at support email. All of the actual help came from people like myself, giving specific advice on known problems.
As the criticism mounted in the forums, the forum admins quietly turned the switches on posts, first not allowing posts with external links, and then later hiding many posts, as they did for a great many of mine on May 30th:
I was completely banned from posting. The forums eventually expired, without them having provided a single item of help.
Meanwhile, over on their official page on Facebook, much the same was happening. Their official updates were posted here as well as on Kickstarter, and although they were much more active here, they again failed to provide any meaningful substance in their replies. And again, as the criticism mounted, heavy censorship ensued, as we can see here:
Once again, their default stance was to contact customer support, which would invariably give them the run around.
This scenario would play itself out again, on Instagram. This time, it was all promotional, and when they started posting stock coffee photos, pictures of the original prototype machine and lattes (which it definitely doesn’t make), it was obvious they had hired a 3rd party contractor to post material.
Once again the criticism mounted, and the person in question begged with customers to contact support and tried to help. However, it was evident that the guy, now revealed to me as Jestan Dale Mendame, based in the Philippines, had bought Orenda’s over-optimistic outlook, hook, line and sinker. When more and more facets of Orenda’s deception were mentioned to him, he doubled down. Eventually, after ignoring requests to stop posting inaccurate information, I revealed his name on Facebook on the unofficial page.
His response to this was decidedly angry, and he insisted he was quitting from helping us or working for Orenda. Promotional material continues to be posted to Instagram to his day (January 2019), so it’s unclear if it’s him or some other sucker.
A few weeks later, censorship again started, and no one was allowed to comment on any of their posts on Instagram. Yet more censorship would follow in January 2019 on the SOSV Facebook page, as mentioned below.
Coffee Makers with Grinders
The history of coffee makers with built in grinders is not very good. In office settings, at least in the United States, Jura seems to dominate, but they are expensive machines designed for high volume, and not home settings. And even such machines are prone to break down, being overly complicated, as witnessed by the kitchen in my work environment, which has been a graveyard for coffee machines of different types.
Historically, the problems with a grinder in the same space as the brewer is moisture. Steam of course rises into exactly the space where the grounds are, which makes them expand and clog things up, to name but one of the problems. I had a discussion with a CEO of a small company that really makes coffee machines, and he expressed concern about such combined machines. In the case on the Orenda, the grinder was one of the attractions to me, but in hindsight, it’s added to an already overly complex machine, with more things to go wrong.
That said, in 2019, there are plenty of such machines. It may be that the designers have worked out problems with past machines.
After months of no real communication, Orenda sent out the following email to many (but not all) customers. I ended up seeing it posted on Facebook.
Our intention for starting Orenda is to bring craft coffee to every home and reduce waste from popular alternatives. While there are people who are having positive experiences with the device that we’ve created, we are not happy that you are not having the same positive experience.
Admittedly, we were inexperienced when we started this and made mistakes along the way. We could keep listing problems here, or even choose the easy way out. Instead, we would like to focus our energy on finding solutions. As we start the new year, our number one goal is to make this right. And we need your help to set this straight.
While many factors were out of our control, one thing we realized we could do better is to get more funding to solve problems and serve you better. The reason we couldn’t do this sooner was because the money from each order has already been used for producing the unit. We have started the fundraising process and have investors coming in. As soon as we finish this round, our number one priority is to ensure that you have a positive experience. With more resources, we are expecting the app experience to continuously evolve, refunds being issued, our spare parts taken care of, our customer service more responsive, QAQC done to the highest standard, shipping and returns promptly processed — just to name a few.
Moving forward, we are looking to complete this fundraising process as soon as possible so we can go ahead and solve those problems for you. We are aiming to finish this within the next 2–3 months. You can expect to hear from us as soon as the process is complete. Until then, we would like to ask for your kind understanding and patience. Please accept our apology for the inconvenience caused. We will be seeing this through all the way and get this right.
If you would like to discuss this update further, we will also be available for a call. Please book our time here at your convenience.
Wish you all the best for the new year.
Nicha, Bank & the Orenda Team
In this case, “Nicha” is Ornicha Srimolka and “Bank” is Pawin Bank Wongtada.
All true, as far as it goes, but what has struck me about their communications, and continues is the superficiality of it all — there’s no real discussion of details, no real promises and much avoidance of real problems. It’s the worst kind of customer service. More on this in a moment.
Call 12th January 2019
I scheduled my call for 6.30am on a Saturday morning. Yes, I’m up then, and yes, I’ve had my coffee from my $20 Black and Decker maker.
Pawin, one of the “founders” called. He didn’t initially have a huge amount to say, apart from reiterating what was in the email (which I didn’t get until much later that same day), so I led the charge. I gave him a short version of the accusations to date (well, as many as I could squeeze into 10 minutes).
Pawin tried to express to me some of his despair and sorrow for the situation — or at least he claimed so. To me it was more an overzealous twenty-something who was going through the motions.
Among other things, I explained that they had been repeatedly dishonest to customers. Not only that, but that I had personally explained issues and found far more problems in my own exploration than they’d ever revealed to customers, and that nothing less than 100% refunds to all customers would be satisfactory. I eventually extracted a promise from him that this would in fact happen when funding occurred, but I also expressed doubt that “2–3 months” was likely, and that any investor would even take this on.
He told me also that they were hiring an app team to take care of software issues. I made very clear to him that in fact most of the software problems were firmware, and that they had mislead the customers over this matter. It didn’t occur to me after to ask how they were going to hire such a team — would this happen after the funding was secured and before the refunds? What is the priority?
He finally promised to chase up the matter of fake FedEx return instructions, which he claimed he was entirely unaware. This is pretty bizarre in an of itself — someone was responsible after all — who? Either someone was incredibly or grossly incompetent or someone was instructed, or took their own initiative to lie to myself (and others). Perhaps we’ll never know.
He finally revealed that their CTO passed away from cancer in 2018. Tragic, certainly, but well after the fact of the initial failures, and only explains a little of the unfortunate events.
After the Call
After some thinking about the call, I came to some conclusions. First, my bet is that the refunds don’t happen, and Orenda simply fail to obtain the funding. The whole deal just doesn’t make sense when appealing to investors. This just isn’t the way that tech investment works.
Also, once again calling all disaffected customers seems an awful and ridiculous waste of time. From a business perspective, directly addressing concerns in an email (which he didn’t initially) would have been far better use of time, as would have been doing this at each stage.
In a follow-up email to Pawin, I highlighted things he had promised, in particular provide refunds in “2–3 months”. He almost immediately reneged on that:
We promise to keep our customers updated with funding. I would also emphasize that we are ‘aiming’ to get funding in 2–3 months, and already have one of the founders dedicated full time to that. Since you work in the Valley, I’m sure you know we cannot control the process that precisely since the deal is not entirely in our control. However, I guarantee we are pushing as hard as we can there and already have a lead.
Making things right with our customers that asked for refunds and existing users is our #1 priority. Ideally, we get the full amount in one shot so we can both process full refunds to everyone that asked, and hire a bigger team to improve the experience for existing users at the same time. In the event that we get funding in smaller chunks, we will be providing refunds to some customers first — as much as the amount allows for. Next chunks will then be spent on more refunds and hiring the dev team.
So the message here has subtly, but conveniently, changed — they will refund when and if they can.
As aside, Pawin informed me that someone claiming to be me made a ton of bookings on their system:
Let’s just say I know how to spell my own name, and I don’t look kindly on impersonation; and even though I know who this probably is, I’m not in a hurry to pass that onto Orenda. But the irony of flooding them with bookings when customers have received dozens of emails from their support saying nothing should not be lost.
A week later, I was still waiting to hear from Pawin or Donna why I was provided with fake RMA information. But I did get this from Pawin:
On a separate note, may I ask why you decided to comment on SOSV’s page?
Doing so will only delay the process longer since we’ll need to stop looking for more investors to tend to the page on behalf of SOSV.
Looking forward to hearing your thoughts.
This is the most spectacularly ironic thing I’ve seen all year (OK, it’s still January). Yes, several customers had started in December posting on SOSV’s Facebook page, asking for refunds, so this was far from a secret. But “delay” after leading customers around for over a year? And what exactly was this “tending” — surely no more censorship; I pointed out to him that would be a shameful things to do.
But on Friday 18th January, it turns out that is exactly what happened — the SOSV page had been “tended” to, and all the comments removed and people banned from posting, making a total of 4 places where dissent had been shutdown.
So what do I conclude from all this January activity? In short, nothing. The playbook does not seem to have changed in the slightest — it’s the same go around with customer support, there’s no addressing of practical issues (e.g, actual discussion of software issues, or ways to address hardware problems), the censorship continues. I think that the Orenda team, such as it is, has spun itself an alternative version of reality about what they’ve done, and no one is likely to shake them free of it, even after a crash and burn failure — which is surely the next thing to come.
But I’ll tell you something else. This drama is far from over.
Pawin Bank Wongtada aka “Bank” — one of the “founders”
Ornicha Srimolka aka “Nicha” — one of the “founders”
Rayan Al-Shaibani — one of the “founders”. Rayan appears in the support videos.
Cyril Ebersweiler — HAX/SOSV investor
Benjamin Joffe — HAX/SOSV investor
Duncan Turner — HAX Director
Noel Joyce — HAX Design Director
Jurn — Head of production
SP — Supply manager
Sally Wang — Sourcing
Peter Wang — Manufacturing
“Eden”, “Michelle”, “Kenney”, “Donna” — Customer support personalities.
Jestan Dale Mendame — Instagram contractor
Thanks to all the members of the “Auroma/Orenda Customers” for supporting material and hanging in there.
October 2019 Update
After the January 2019 update, Pawin contacted me. After a lot of back and forth trying to get FedEx details, and it becoming blatantly obvious what I had known all along — that the support team was clueless/dishonest — I finally got an actual FedEx shipping label. Some weeks after, after more badgering indicating I had in fact returned it, I got my $434 back! Amazing. At least one other person got at partial refund. It’s hard to know how many other people did — we can safely assume it was a very small number.
But there was a final twist here — you’ll recall I had obtained a second machine from a generous person (whose details I no longer have), and it was in fact their machine I returned. This was a machine which I hadn’t modified, and therefore was arguably still under warranty. Clearly they weren’t checking serial numbers on this.
I indicated to Pawin that I had another machine I wanted a refund for (had I got such a refund, I would have gladly made an attempt to find the owner of the other), but there was no response, and indeed, this seems to have been the last anyone heard of Pawin. I ended up taking the machine to the local recycling center; and will claim a small tax break on it. Oh well.
Finally, for those of you who contacted me about further development on the firmware — sorry, I quite literally wrote down 100% of everything I found out about the machine. So, whilst I appreciate your emails, I simply have nothing to add. But good luck.