Why Robots Can Backflip but Struggle With Real Work
Enzo Wälchli · Operator Circle
EP. 13
Why Your First Robot Project Should Start Small | Andie Zhang, ABB Robotics
Andie Zhang of ABB Robotics on collaborative robots, the gap between impressive demos and useful automation, and why the first deployment should start with a small, well-defined task.
Discuss this episode and meet other Swiss tech builders in the TechDrive WhatsApp group.
Key insights: executive summary
- Start with a small job. Andie advises first-time robot users to break a person’s work into individual tasks, then choose a manageable part to automate. The first project also builds the company’s skills in deployment and operator training.
- Plan the work around the robot. A reliable machine still needs materials, usable interfaces and a sensible process. Andie explains why replenishment and the surrounding workflow can determine whether automation delivers the expected benefit.
- A backflip does not prove practical autonomy. She welcomes the attention humanoids bring to robotics, while asking companies to be clear about what their machines actually do autonomously. An impressive movement does not establish an ability to perform an unrelated everyday task.
- Live demos expose hidden assumptions. A YuMi bracelet demonstration became a dance after an all-night rehearsal failed to reach the reliability the team needed. Another demo struggled under reporters’ lighting. Andie says the bracelet application eventually worked.
- Build and test with people beyond the usual robotics CV. Andie connects broader hiring, diverse teams and hands-on user testing with products that more people can use. She also asks hiring managers to check whether their job requirements unnecessarily narrow the applicant pool.
- Features need a customer problem. Drawing on her experience at Tetra Pak, Andie explains value-based selling: understand the customer’s need, identify how the product addresses it and make that value clear. A longer feature list alone does not justify a higher price.
Start with a small task, then build the experience to automate more.
A product feature creates value when it solves a customer’s problem.
How should a company choose its first robot project?
Andie sees first-time users make an ambitious assumption: an advanced robot should be given the hardest job. Her advice is to begin by observing the person doing the work. What looks like one job may contain many separate tasks, only some of which are suitable for automation.
She suggests starting with a small, well-defined task that is repetitive, physically demanding or requires consistent precision. The first deployment takes time because the company is also learning to train operators and organise the work. That experience can make later deployments easier.
Why does cobot adoption depend on more than the robot?
The conversation moves from the early emphasis on collaborative-robot safety to the practical reasons customers choose cobots: compact equipment, approachable programming and easier integration into an existing workplace.
Andie also asks teams to examine the surrounding process. Who replenishes the materials? Can the robot keep working, or does someone need to interrupt another job to feed it? A deployment can require changes to the workflow as well as a capable machine. Ease of use is part of the product’s value, not a detail to consider after installation.
What do robot backflips tell us about physical AI?
Humanoid demonstrations have brought robotics to a wider audience, which Andie sees as useful for the industry. Her concern is the inference viewers may make: seeing a backflip does not establish that a machine can load a dishwasher or perform another useful task autonomously.
She contrasts that spectacle with a practical example of physical AI: machine vision helping an industrial arm decide how to pick unfamiliar bags of clothing for warehouse handling. The result may be less striking on social media, but it addresses a specific task. Her request to robotics companies is to communicate capabilities and autonomy clearly.
What happened when the YuMi bracelet demo went wrong?
For a launch in Tokyo, the team planned a bracelet-making demonstration involving two single-arm YuMi robots and a dual-arm YuMi. Delicate beads made failures particularly painful: when something went wrong, the team had to collect the scattered pieces and start again.
After working through the night, they switched to a robot dance for the launch. Andie says they later got the bracelet demonstration working. She also recalls a separate occasion when reporters’ lighting interfered with the cameras. Both stories explain why she wants to understand a system’s details and see it work live before accepting an ambitious product claim.
How can diverse teams make robots easier to use?
Andie describes asking a new communications colleague, who had no robotics background, to set up a single-arm YuMi using its instructions. The colleague needed help but completed the task, giving the team a useful view of the experience beyond its own expertise.
She connects this with hiring: requirements such as years in robotics can exclude people with relevant skills and the ability to learn. She also draws on work in China and Europe to explain why cultural context matters. Testing with people of different backgrounds, heights, hand sizes and visual needs can reveal assumptions that a uniform team misses.
How do you turn robot features into customer value?
At Tetra Pak, Andie worked on a global value-based-selling initiative. She still applies its central question in robotics: what problem does this customer have, and how does the product solve it?
A feature matters commercially when its benefit is relevant to that customer. Andie connects this to product management and startup decisions: identify the target customer, understand the actual need and check whether the proposed solution is overengineered or insufficient. The work starts before the sales pitch, when the team decides what to build.
Who is Andie Zhang?
Andie Zhang is Global Portfolio and Technology Manager Collaborative Robots at ABB Robotics. In this interview, she describes joining ABB Robotics in September 2016 as its first global product manager responsible for collaborative robots, after ten years at Tetra Pak. Born in China and raised in Sweden, she also shares how working across China and Europe has shaped her approach to teams, product development and customers. The conversation brings together her experience with cobot strategy, deployment, user testing and value-based selling.
Episode show notes
A robot can do a backflip. That does not tell you whether it can load a dishwasher or solve a customer’s problem.
Andie Zhang joins Justinas Mišeikis for a drive through Zürich about the practical work behind useful robots. From a YuMi launch demo that became a dance to a newcomer setting up a robot, the conversation explores first deployments, workflow design, responsible product claims, hiring and user testing. Andie also shares the value-based-selling lesson she brought from Tetra Pak into robotics: start with the customer’s problem.
More conversations from the Swiss tech ecosystem
Nobody Has Left a Robot Working Overnight
Yulia Sandamirskaya · ZHAW Zurich University of Applied Sciences
The next drive, in your inbox.
A short note when a new episode drops: the guest, the hook, and the one question you won't want to miss. No spam.
Follow the Swiss tech ecosystem, one drive at a time
New in-car interviews with founders, investors and researchers building Zurich's deep tech, AI and robotics scene. Want to discuss this episode, ask follow-up questions and meet other builders?