Orchard POS, OS Is Best Tech We’ve Seen in This Industry. Seriously.

Three years ago two Dartmouth students, Thomas Cecil and Ian Bateman, reached out to us with the idea of building yet another POS system. They had already built and sold a very successful online ordering company while in their junior year of undergrad (it grew to $400,000 in revenue in under nine months) and throughout the process of building that they discovered the dumpster fire that is POS.

Despite being only 20 years old, their observations were more accurate than industry “professionals”:

1. Legacy and even emerging cloud POS companies had horrible software practices
2. Integrations into POS systems were an absolute nightmare
3. POS companies cared more about their payments processing residuals than they did the quality of the POS product

While we congratulated them on recognizing more about this industry in nine months than merchants who had built their entire careers here, we advised them against building POS for one simple reason:

In brick and mortar, distribution is king. And they had none.

Yet that didn’t dissuade them. If anything, it only made them more recalcitrant.

Three years later the pair emerged with Orchard Systems, a POS and operating system (OS) that merit serious dialogue for the implications it brings not only to the POS and payments industries, but embedded systems across the globe.

Here’s what happened:

After ignoring us, Orchard built their POS and targeted enterprise quick service and fast casual operators. Throughout their sales process they kept hearing the same objection:

“Your software is head and shoulders above what we have but we can’t afford to swap out our hardware; on a per-store basis we’re talking about a $30,000, $40,000 investment that we’re not prepared to make.”

For the record, that quote is paraphrased from one of McDonald’s most influential franchisees.

Well, like our conversations with them a year earlier, Orchard was not to be dissuaded. Orchard did what any good engineer does and buckled down to solve the problem.

Any holy crap did they solve the problem.

Orchard hired the lead engineers that built Chrome OS for Google, the founders and maintainers of Hardened Gentoo Linux (which is more colloquially known as the most secure Linux distribution in existence), and swaths of engineers that built their own and/or maintain popular programming languages (including LASP, Elixir, Dartlang, and a bunch of stuff this industry wouldn’t understand so we won’t waste our text). In other words, their problem was so hard that they were able to attract literally the highest caliber engineering talent we’ve ever seen in POS and payments. (Note: really smart people like to work on really hard problems because solving puzzles is fun; if you’re not an engineer you likely won’t understand this.)

What emerged was an OS that could nuke any piece of ARM or x86 hardware and allow Orchard to run their POS app… or really any other app, for that matter. In simple terms, a merchant doesn’t want to move POS systems because they’re strapped to that $30K Micros hardware that’s “locked down” to run Micros software only?

Not anymore. <- watch this video

We’ve seen Orchard’s hardware room – which is effectively a room full of graveyard business models now that Orchard can nuke them all – and it has more than 20 POS devices, from Toast and Micros to NCR and PAR. Some of the hardware devices came out 15 years ago and are just as easily wiped clean.

In addition to the nuking capability, we believe Orchard has built an OS that is analogous to what Amazon did with cloud services a decade ago. In the early 2000’s you’d hire a team of IT people to plug cables into a data center and take calls around the clock if your server went down. Then this thing called AWS showed up, boasting 99.99999% uptime, and it became a no-brainer to license servers from someone else. Cloud-as-a-Service.

Orchard is likewise OS-as-a-Service, simplifying the cost and IT needs to manage an OS.

Orchard OS is a distribution (distro) of Linux. Most people don’t know this, but Linux is inherently free. It’s also purpose-built, which only further reduces its cost. On Windows or Android, OSs built for personal computing and smartphones respectively, you’re paying for features (and complexity) you won’t need in a POS environment. Think of it like cable: under the traditional model of the 1990’s you’re paying for a ton of channels that you’ll never watch. This has given rise to unbundling, where you only pay for the channels you want to view. Why pay for the bloat of a general purpose OS which, by the way, also makes POS devices run slower?

When you use a bloated OS for POS purposes you also drastically increase your security risk. For instance, if you’re running an iOS or Android device, why would you want to allow your employees to download other apps or browse the web while at work? Not only does this distract from the reason you pay them, it exposes vulnerabilities where some moron gets a virus on your critical business machine (i.e. POS) from jerking around on Farmville. Orchard’s OS doesn’t allow this to happen. As payments companies consume the entire POS industry they should be aware that they’re taking on this level of subpar security (although to be fair, payment cradles are also pathetically insecure because software companies aren’t building this software – see Arby’s breach).

There are also massive strategic benefits to using Orchard’s OS. When you run on a smartphone OS (which is precisely what iOS and Android are) you’re held hostage to their rules. In choosing Android a user is either on the perpetual treadmill to keep up with Android’s updates or they fork a version of Android and maintain it locally, falling way behind the open source community and accruing massive technical debt (this latter approach is what Toast has done and we don’t think it will prove sustainable).

Or you risk that any app store unilaterally changes the rules for acceptable apps and screws up your entire business model. Giving the argument some teeth, in early 2018 Apple forbid the publication of reskinned apps, where one app could serve many users by simply changing the logo (think of companies like LevelUp and ChowNow). You must also deal with the app store policies that can restrict or severely delay updates. Found a critical bug and need to hot code swap for a quick fix? App stores don’t allow this feature in published apps, and it could take weeks for an app store to approve your update because you have to resubmit the entire application for review.

If you’re a techie, here are other benefits of Orchard’s OS: A/B root swap for the operating system, device health monitoring, full (but secure) ssh shell access, ongoing managed kernel-level security updates, backwards (legacy) hardware compatibility, automatic application respawn on crash, programmatic VPNs (custom defined device clusters), P2P file syncing across clustered devices, and the ability to simultaneously execute scripts across custom-defined groups of devices.

There’s also hardware metrics monitoring (CPU, RAM, disk usage, etc.), comprehensive system logging (all POS actions/data are logged in real time), online/offline status monitoring, a unique device identification utility (unspoofable – useful for ensuring accurate license billing/reporting), OS package/firmware & application automatic rollback mechanisms, and randomized/targeted canary deployments for production.

If the OS wasn’t impressive enough, keep in mind that Orchard also built an entire restaurant software suite (POS, kiosks, KDS, digital signage, timekeeping, online ordering dashboard, and a real time back office management web app) that is legitimately the best-architected system we’ve ever seen.

As evidence of their pace of innovation, Orchard engineers were able to build a KDS system using commodity hardware (Raspberry Pi) that reduced the end-cost to merchants to less than $40. They took this approach when MicroPlus (who does KDS for large QSR’s like CookOut and Bojangles) told them that their KDS would cost upwards of $700, not including the display hardware. “In an industry as cost-conscious as hospitality, your margin is our opportunity,” laughs Ian Bateman, Orchard’s CTO and cofounder.

Part of Orchard’s POS success is undoubtedly attributable to the fact that Orchard doesn’t have to deal with old technologies; the other part is because they’re smart enough engineers not to choose those tools to begin with.

Most legacy POS systems operated in a master-slave relationship: one terminal onsite was the designated “master” and the other terminals its “slaves”. When the master died, the slaves were useless and the business would grind to a halt. Smarter engineers realized this was a huge liability and built what’s called peer-to-peer (P2P) computing. In a P2P environment, tasks are shared across a network of devices, eliminating a single point of failure.

Orchard not only implemented P2P architecture (there are other POS companies that have done this too, by the way), but they’ve eliminated the idea of on-premise servers.

Orchard’s robust backend consists of traditional cloud servers coupled with edge-computing proxy-servers in each POS device in the store. Data is stored in the proxy layer and synced to the Orchard cloud when a device is connected to the internet. This syncing is instantaneous and uses advanced CRDT algorithms to mathematically ensure no data duplication occurs despite unreliable internet connection. When there is no internet connection, business logic data spools into the proxy layer and is stored, encrypted, on a disk until reconnection occurs, at which point all locally stored data is synced to the servers and then deleted from the terminal unless kept there intentionally.

Their POS devices maintains complete offline functionally despite no internet connection. This means checks and other business data are still replicated across each POS device using CRDTs. These were architected for Orchard by Dr. Christopher Meiklejohn, the inventor of LASP and an expert in CRDT technology.

Because Orchard OS comes bundled with a software-defined VPN, they can also associate and group different devices with a store or store group. This allows Orchard to cleanly manage all stores in a brand by a tree structure hierarchy. Accordingly, Orchard’s back office software can mimic your restaurant brand’s corporate hierarchy, like:

Corporate office → Southeast → North Carolina → Raleigh → Store #1235

and apply business logic (menus, integrations, prices, etc.) at any level in the tree and that business logic will cascade (i.e. apply itself to lower nodes in the tree). For example, you can apply a National Menu (think Dollar menu) at the Corporate office node, and all stores will instantly have that menu. Nodes below can apply business logic as well – like regional promotions or price changes.

Furthermore, the software uses an intelligent permissions structure so that a lower node (e.g. Store #1235) cannot override the business logic set by a higher node (e.g. corporate office) unless the higher node allows it to. This is critical for franchise systems (like McDonald’s) where the franchisor needs to maintain consistency across all stores.

If that wasn’t impressive enough, Orchard has an open and documented API. Partners can use it in a RESTful fashion, but it’s technically a GraphQL API, a query language that was developed by Facebook to make it easier, more cost effective, and faster to pull & push data at scale. All parts of Orchard’s POS system are accessible via this API as well, which means integrating 3rd party solutions (online ordering, loyalty, etc.) is much faster and easier.

While Orchard has limited itself to QSR and fast casual today, features for table service are a quarter away. “Features are a known quantity,” explained Thomas Cecil, Orchard’s CEO and cofounder. “In the enterprise segment you can get sucked into custom feature hell. But for 99% of the market it’s the classic 80/20. And it doesn’t matter how good your features are if your POS doesn’t even work.”

We strongly agree with this; look at Aloha POS as an example of a POS with robust features but a business model and core product that is irrelevant in today’s POS market. “Aloha hasn’t produced a cloud POS replacement in what, a decade? That shows you how hard it is to rebuild architecture,” Thomas points out.

Orchard has found many ISOs and banks lining up to license their POS software for resale. “When MobileBytes sold to Heartland a lot of their reseller-ISOs found the new processing economics unworkable. Several have come to us since we don’t give a rip about the processing revenue share,” says Thomas.

We wouldn’t be surprised if ISOs would prefer to eliminate POS risk altogether and just buy Orchard. Given the potential in Orchard’s OS business we’ve advised them to do just this – sell the POS business and focus on licensing the operating system. Even the online ordering companies, who have a burgeoning number of tablets on merchant countertops, can now use Orchard OS to turn those tablets into a full-blown POS.

“Yeah, I guess we’re finally listening to you and agree we aren’t going to build distribution for POS ourselves. So we’re selling the POS. And we are 99% sure it’s the best-engineered restaurant POS in the industry.”

As Orchard’s most formal advisors, we have to agree.

For more questions about Orchard, reach out to Thomas [at] Orchardsystems.com