Hubert Wang

I am
Hubert Wang

Wechat Official Account
Find fun things here!

Artly Robot Barista Coffee Experience

Good experience, good coffee!


Advertising board and store front

Interestingly, it looks like a "unicorn" company from the VI design haha. I noticed many new unicorn companies have similar design styles, e.g. Robinhood, Plaid. They use thin stroke with high-saturated color. More details on their VI design can be found here.

Order by pad

I ordered by the pad in the store this time. I can also order in App using my phone ahead to pick up next time to save time.

Barista bot working in progress to make a latte

Artly uses a bot arm to mimic the process of a barista. Several interesting things:

  • It is a one-arm barista. It took around 3 minutes (ref to the photo below) to make a 12oz latte. In comparison, a Starbucks barista only needs 36 seconds even for one of the most complex drinks. And of course every second counts for a coffee store.
  • Some steps used workaround approaches. An example is wiping the steam stick. I was wondering how the bot arm can wipe the remaining milk on the steam stick after frothing milk (It is not easy even for a human being because the milk solidifies quickly with high temp, and usually a barista needs a clean wet mop to clean it). It turned out the company invented a moving circle part around the steam stick to flush milk away with warm water flow.
  • Some manual steps are still needed, e.g. refilling the bean hopper. There was one storekeeper when I took the video.
  • Latte Art! Wow

The overall experience is better than my original expectation. I'll def revisit to get a fair-price good-quality coffee as a consumer. Some thoughts are listed below for Artly as a start-up new to the coffee market.

Risks as a startup

  • Less efficient than a human being. As mentioned above, 5x less efficiency in making a drink is a huge drawback for investors to consider. There are some mitigators in store e.g. setting up multiple bots in one store to parallelize, but it comes with a tradeoff in set-up cost.
  • Operating costs can be high: replenishment, ingredient preparation, recovery fee from failures, machine maintenance need experts, people for store watching (savable?)
  • Complex failure models: A bot with more sensors have a higher probability to fail and is harder to fix. In Artly's case, the whole system also includes software (control algorithms, app, name projections, etc.) which can fail in multiple ways. It would be super interesting to test how this system can fail (if there is no keeper in store ;) ), e.g. stress testing with many orders in the queue; cases running out of beans, milk, cups; unstandard cups; dependent devices like cup washer break down, etc.
  • Store scalability is limited by the high cost of set-up: The robot is expensive (initial cost and operational cost mentioned above). Expanding this mode quickly is a pain point for the bot barista business.
  • Missing communication for Human-centered design. No communication between human beings and robots. Benchmarking: Starbucks is actually famous for its connection, coffee as a service.


  • Varied types of drinks. The coffee chain has a higher cost to change or expand their menu than a bot coffee shop due to the high educational cost for human beings. Bot coffee shops can send users new types of drinks on daily basis to increase the revisit-rate. Risk is the increased cost and difficulty of ingredients preparation. Artly is already trying this, you can tell from the over 100 types of drinks on its menu.
  • High-end coffee market. The first time I saw bot coffee was in CES 2018, in a manufacturer's showcase a bot (actually just a movable water tap) can mimic the awardee wining champion’s hand-pouring curves to democrat high-end coffee-making knowledge. This path is attractive for coffee lovers, to pay extra to get really good hand-pouring specialty coffee. This is a different path than the one Artly currently chooses. A branch brand trying this out may be worth it: Artly is hard to beat Starbucks which has been optimized crazily on every step that a coffee chain needs hence hard to beat, but Philz and Blue bottle are much easier to beat and it can be a good strategy to start with high-end branding and then as a lower-leveled attack to Strarbucks after having a good brand awareness.
  • ERP at an extremely low level (by cup). With sensor integration, data can be collected at the cup level to plan resources at an incredibly precise level. This can potentially reduce waste hence costs. Data enables a way more precise resource planning than competitors. Personalized preferences learnings to enable better customer experience.

I personally hope to see Artly goes into the 2nd bullet point as a coffee lover so I can try all coffee masters' drinks. I'm willing to even pay for the machine to get one at home if it can provide many champions' pouring over with OTA updates.

Write a Comment